chore(evals): rollback prompt-steering and behavioral changes to main

Co-authored-by: Sehoon Shon <sshon@google.com>

.gemini/commands/fix-behavioral-eval.toml

@@ -1,60 +0,0 @@
-description = "Check status of nightly evals, fix failures for key models, and re-run."
-prompt = """
-You are an expert at fixing behavioral evaluations.
-
-1. **Investigate**:
-   - Use 'gh' cli to fetch the results from the latest run from the main branch: https://github.com/google-gemini/gemini-cli/actions/workflows/evals-nightly.yml.
-   - DO NOT push any changes or start any runs. The rest of your evaluation will be local.
-   - Evals are in evals/ directory and are documented by evals/README.md.
-   - The test case trajectory logs will be logged to evals/logs.
-   - You should also enable and review the verbose agent logs by setting the GEMINI_DEBUG_LOG_FILE environment variable.
-   - Identify the relevant test. Confine your investigation and validation to just this test.
-   - Proactively add logging that will aid in gathering information or validating your hypotheses.
-
-2. **Fix**:
-   - If a relevant test is failing, locate the test file and the corresponding prompt/code.
-   - It's often helpful to make an extreme, brute force change to see if you are changing the right place to make an improvement and then scope it back iteratively.
-   - Your **final** change should be **minimal and targeted**.
-   - Keep in mind the following:
-     - The prompt has multiple configurations and pieces. Take care that your changes
-       end up in the final prompt for the selected model and configuration.
-     - The prompt chosen for the eval is intentional. It's often vague or indirect
-       to see how the agent performs with ambiguous instructions. Changing it should
-       be a last resort.
-     - When changing the test prompt, carefully consider whether the prompt still tests
-       the same scenario. We don't want to lose test fidelity by making the prompts too
-       direct (i.e.: easy).
-     - Your primary mechanism for improving the agent's behavior is to make changes to
-       tool instructions, system prompt (snippets.ts), and/or modules that contribute to the prompt.
-     - If prompt and description changes are unsuccessful, use logs and debugging to
-       confirm that everything is working as expected.
-    - If unable to fix the test, you can make recommendations for architecture changes
-      that might help stablize the test. Be sure to THINK DEEPLY if offering architecture guidance.
-      Some facts that might help with this are:
-      - Agents may be composed of one or more agent loops.
-      - AgentLoop == 'context + toolset + prompt'. Subagents are one type of agent loop.
-      - Agent loops perform better when:
-        - They have direct, unambiguous, and non-contradictory prompts.
-        - They have fewer irrelevant tools.
-        - They have fewer goals or steps to perform.
-        - They have less low value or irrelevant context.
-      - You may suggest compositions of existing primitives, like subagents, or
-        propose a new one.
-      - These recommendations should be high confidence and should be grounded
-        in observed deficient behaviors rather than just parroting the facts above.
-        Investigate as needed to ground your recommendations.
-
-3. **Verify**:
-   - Run just that one test if needed to validate that it is fixed. Be sure to run vitest in non-interactive mode.
-   - Running the tests can take a long time, so consider whether you can diagnose via other means or log diagnostics before committing the time. You must minimize the number of test runs needed to diagnose the failure.
-   - After the test completes, check whether it seems to have improved.
-   - You will need to run the test 3 times for Gemini 3.0, Gemini 3 flash, and Gemini 2.5 pro to ensure that it is truly stable. Run these runs in parallel, using scripts if needed.
-   - Some flakiness is expected; if it looks like a transient issue or the test is inherently unstable but passes 2/3 times, you might decide it cannot be improved.
-
-4. **Report**:
-   - Provide a summary of the test success rate for each of the tested models.
-   - Success rate is calculated based on 3 runs per model (e.g., 3/3 = 100%).
-   - If you couldn't fix it due to persistent flakiness, explain why.
-
-{{args}}
-"""

.gemini/commands/promote-behavioral-eval.toml

@@ -1,29 +0,0 @@
-description = "Promote behavioral evals that have a 100% success rate over the last 7 nightly runs."
-prompt = """
-You are an expert at analyzing and promoting behavioral evaluations.
-
-1. **Investigate**:
-   - Use 'gh' cli to fetch the results from the most recent run from the main branch: https://github.com/google-gemini/gemini-cli/actions/workflows/evals-nightly.yml.
-   - DO NOT push any changes or start any runs. The rest of your evaluation will be local.
-   - Evals are in evals/ directory and are documented by evals/README.md.
-   - Identify tests that have passed 100% of the time for ALL enabled models across the past 7 runs in a row.
-   - NOTE: the results summary from the most recent run contains the last 7 runs test results. 100% means the test passed 3/3 times for that model and run.
-   - If a test meets this criteria, it is a candidate for promotion.
-
-2. **Promote**:
-   - For each candidate test, locate the test file in the evals/ directory.
-   - Promote the test according to the project's standard promotion process (e.g., moving it to a stable suite, updating its tags, or removing skip/flaky annotations). 
-   - Ensure you follow any guidelines in evals/README.md for stable tests.
-   - Your **final** change should be **minimal and targeted** to just promoting the test status.
-
-3. **Verify**:
-   - Run the promoted tests locally to validate that they still execute correctly. Be sure to run vitest in non-interactive mode.
-   - Check that the test is now part of the expected standard or stable test suites.
-
-4. **Report**:
-   - Provide a summary of the tests that were promoted.
-   - Include the success rate evidence (7/7 runs passed for all models) for each promoted test.
-   - If no tests met the criteria for promotion, clearly state that and summarize the closest candidates.
-
-{{args}}
-"""

.gemini/settings.json

@@ -2,7 +2,8 @@
   "experimental": {
     "plan": true,
     "extensionReloading": true,
-    "modelSteering": true
+    "modelSteering": true,
+    "memoryManager": true
   },
   "general": {
     "devtools": true

.gemini/skills/behavioral-evals/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: behavioral-evals
+description: Guidance for creating, running, fixing, and promoting behavioral evaluations. Use when verifying agent decision logic, debugging failures, debugging prompt steering, or adding workspace regression tests.
+---
+
+# Behavioral Evals
+
+## Overview
+
+Behavioral evaluations (evals) are tests that validate the **agent's decision-making** (e.g., tool choice) rather than pure functionality. They are critical for verifying prompt changes, debugging steerability, and preventing regressions.
+
+> [!NOTE]
+> **Single Source of Truth**: For core concepts, policies, running tests, and general best practices, always refer to **[evals/README.md](file:///Users/abhipatel/code/gemini-cli/docs/evals/README.md)**.
+
+---
+
+## 🔄 Workflow Decision Tree
+
+1.  **Does a prompt/tool change need validation?**
+    *   *No* -> Normal integration tests.
+    *   *Yes* -> Continue below.
+2.  **Is it UI/Interaction heavy?**
+    *   *Yes* -> Use `appEvalTest` (`AppRig`). See **[creating.md](references/creating.md)**.
+    *   *No* -> Use `evalTest` (`TestRig`). See **[creating.md](references/creating.md)**.
+3.  **Is it a new test?**
+    *   *Yes* -> Set policy to `USUALLY_PASSES`.
+    *   *No* -> `ALWAYS_PASSES` (locks in regression).
+4.  **Are you fixing a failure or promoting a test?**
+    *   *Fixing* -> See **[fixing.md](references/fixing.md)**.
+    *   *Promoting* -> See **[promoting.md](references/promoting.md)**.
+
+---
+
+## 📋 Quick Checklist
+
+### 1. Setup Workspace
+Seed the workspace with necessary files using the `files` object to simulate a realistic scenario (e.g., NodeJS project with `package.json`).
+*   *Details in **[creating.md](references/creating.md)***
+
+### 2. Write Assertions
+Audit agent decisions using `rig.setBreakpoint()` (AppRig only) or index verification on `rig.readToolLogs()`.
+*   *Details in **[creating.md](references/creating.md)***
+
+### 3. Verify
+Run single tests locally with Vitest. Confirm stability locally before relying on CI workflows.
+*   *See **[evals/README.md](file:///Users/abhipatel/code/gemini-cli/docs/evals/README.md)** for running commands.*
+
+---
+
+## 📦 Bundled Resources
+
+Detailed procedural guides:
+*   **[creating.md](references/creating.md)**: Assertion strategies, Rig selection, Mock MCPs.
+*   **[fixing.md](references/fixing.md)**: Step-by-step automated investigation, architecture diagnosis guidelines.
+*   **[promoting.md](references/promoting.md)**: Candidate identification criteria and threshold guidelines.
+

.gemini/skills/behavioral-evals/assets/interactive_eval.ts.txt

@@ -0,0 +1,27 @@
+import { describe, expect } from 'vitest';
+import { appEvalTest } from './app-test-helper.js';
+
+describe('interactive_feature', () => {
+  // New tests MUST start as USUALLY_PASSES
+  appEvalTest('USUALLY_PASSES', {
+    name: 'should pause for user confirmation',
+    files: {
+      'package.json': JSON.stringify({ name: 'app' })
+    },
+    prompt: 'Task description here requiring approval',
+    timeout: 60000, 
+    setup: async (rig) => {
+      // ⚠️ Breakpoints are ONLY safe in appEvalTest
+      rig.setBreakpoint(['ask_user']);
+    },
+    assert: async (rig) => {
+      // 1. Wait for the breakpoint to trigger
+      const confirmation = await rig.waitForPendingConfirmation('ask_user');
+      expect(confirmation).toBeDefined();
+
+      // 2. Resolve it so the test can finish
+      await rig.resolveTool(confirmation);
+      await rig.waitForIdle();
+    },
+  });
+});

.gemini/skills/behavioral-evals/assets/standard_eval.ts.txt

@@ -0,0 +1,30 @@
+import { describe, expect } from 'vitest';
+import { evalTest } from './test-helper.js';
+
+describe('core_feature', () => {
+  // New tests MUST start as USUALLY_PASSES
+  evalTest('USUALLY_PASSES', {
+    name: 'should perform expected agent action',
+    setup: async (rig) => {
+      // For mocking offline MCP:
+      // rig.addMockMcpServer('workspace-server', 'google-workspace');
+    },
+    files: {
+      'src/app.ts': '// some code',
+    },
+    prompt: 'Task description here',
+    timeout: 60000, // 1 minute safety limit
+    assert: async (rig, result) => {
+      // 1. Audit the trajectory (Safe for standard evalTest)
+      const logs = rig.readToolLogs();
+      const hasTool = logs.some((l) => l.toolRequest.name === 'read_file');
+      expect(hasTool, 'Agent should have read the file').toBe(true);
+
+      // 2. Assert efficiency (Cost/Turn)
+      expect(logs.length).toBeLessThan(5);
+
+      // 3. Assert final output
+      expect(result).toContain('Expected Keyword');
+    },
+  });
+});

.gemini/skills/behavioral-evals/references/creating.md

@@ -0,0 +1,151 @@
+# Creating Behavioral Evals
+
+## 🔬 Rig Selection
+
+| Rig Type          | Import From            | Architecture                                                         | Use When                                                                                              |
+| :---------------- | :--------------------- | :------------------------------------------------------------------- | :---------------------------------------------------------------------------------------------------- |
+| **`evalTest`**    | `./test-helper.js`     | **Subprocess**. Runs the CLI in a separate process + waits for exit. | Standard workspace tests. **Do not use `setBreakpoint`**; auditing history (`readToolLogs`) is safer. |
+| **`appEvalTest`** | `./app-test-helper.js` | **In-Process**. Runs directly inside the runner loop.                | UI/Ink rendering. Safe for `setBreakpoint` triggers.                                                  |
+
+---
+
+## 🏗️ Scenario Design
+
+Evals must simulate realistic agent environments to effectively test
+decision-making.
+
+- **Workspace State**: Seed with standard project anchors if testing general
+  capabilities:
+  - `package.json` for NodeJS environments.
+  - Minimal configuration files (`tsconfig.json`, `GEMINI.md`).
+- **Structural Complexity**: Provide enough files to force the agent to _search_
+  or _navigate_, rather than giving the answer directly. Avoid trivial one-file
+  tests unless testing exact prompt steering.
+
+---
+
+## ❌ Fail First Principle
+
+Before asserting a new capability or locking in a fix, **verify that the test
+fails first**.
+
+- It is easy to accidentally write an eval that asserts behaviors that are
+  already met or pass by default.
+- **Process**: reproduce failure with test -> apply fix (prompt/tool) -> verify
+  test passes.
+
+---
+
+## ✋ Testing Patterns
+
+### 1. Breakpoints
+
+Verifies the agent _intends_ to use a tool BEFORE executing it. Useful for
+interactive prompts or safety checks.
+
+```typescript
+// ⚠️ Only works with appEvalTest (AppRig)
+setup: async (rig) => {
+  rig.setBreakpoint(['ask_user']);
+},
+assert: async (rig) => {
+  const confirmation = await rig.waitForPendingConfirmation('ask_user');
+  expect(confirmation).toBeDefined();
+}
+```
+
+### 2. Tool Confirmation Race
+
+When asserting multiple triggers (e.g., "enters plan mode then asks question"):
+
+```typescript
+assert: async (rig) => {
+  let confirmation = await rig.waitForPendingConfirmation([
+    'enter_plan_mode',
+    'ask_user',
+  ]);
+
+  if (confirmation?.name === 'enter_plan_mode') {
+    rig.acceptConfirmation('enter_plan_mode');
+    confirmation = await rig.waitForPendingConfirmation('ask_user');
+  }
+  expect(confirmation?.toolName).toBe('ask_user');
+};
+```
+
+### 3. Audit Tool Logs
+
+Audit exact operations to ensure efficiency (e.g., no redundant reads).
+
+```typescript
+assert: async (rig, result) => {
+  await rig.waitForTelemetryReady();
+  const toolLogs = rig.readToolLogs();
+
+  const writeCall = toolLogs.find(
+    (log) => log.toolRequest.name === 'write_file',
+  );
+  expect(writeCall).toBeDefined();
+};
+```
+
+### 4. Mock MCP Facades
+
+To evaluate tools connected via MCP without hitting live endpoints, load a mock
+server configuration in the `setup` hook.
+
+```typescript
+setup: async (rig) => {
+  rig.addMockMcpServer('workspace-server', 'google-workspace');
+},
+assert: async (rig) => {
+  await rig.waitForTelemetryReady();
+  const toolLogs = rig.readToolLogs();
+  const workspaceCall = toolLogs.find(
+    (log) => log.toolRequest.name === 'mcp_workspace-server_docs.getText'
+  );
+  expect(workspaceCall).toBeDefined();
+};
+```
+
+---
+
+## ⚠️ Safety & Efficiency Guardrails
+
+### 1. Breakpoint Deadlocks
+
+Breakpoints (`setBreakpoint`) pause execution. In standard `evalTest`,
+`rig.run()` waits for the process to exit _before_ assertions run. **This will
+hang indefinitely.**
+
+- **Use Breakpoints** for `appEvalTest` or interactive simulations.
+- **Use Audit Tool Logs** (above) for standard trajectory tests.
+
+### 2. Runaway Timeout
+
+Always set a budget boundary in the `EvalCase` to prevent runaway loops on
+quota:
+
+```typescript
+evalTest('USUALLY_PASSES', {
+  name: '...',
+  timeout: 60000, // 1 minute safety limit
+  // ...
+});
+```
+
+### 3. Efficiency Assertion (Turn limits)
+
+Check if a tool is called _early_ using index checks:
+
+```typescript
+assert: async (rig) => {
+  const toolLogs = rig.readToolLogs();
+  const toolCallIndex = toolLogs.findIndex(
+    (log) => log.toolRequest.name === 'cli_help',
+  );
+
+  expect(toolCallIndex).toBeGreaterThan(-1);
+  expect(toolCallIndex).toBeLessThan(5); // Called within first 5 turns
+};
+```

.gemini/skills/behavioral-evals/references/fixing.md

@@ -0,0 +1,71 @@
+# Fixing Behavioral Evals
+
+Use this guide when asked to debug, troubleshoot, or fix a failing behavioral
+evaluation.
+
+---
+
+## 1. 🔍 Investigate
+
+1.  **Fetch Nightly Results**: Use the `gh` CLI to inspect the latest run from
+    `evals-nightly.yml` if applicable.
+    - _Example view URL_:
+      `https://github.com/google-gemini/gemini-cli/actions/workflows/evals-nightly.yml`
+2.  **Isolate**: DO NOT push changes or start remote runs. Confine investigation
+    to the local workspace.
+3.  **Read Logs**:
+    - Eval logs live in `evals/logs/<test_name>.log`.
+    - Enable verbose debugging via `export GEMINI_DEBUG_LOG_FILE="debug.log"`.
+4.  **Diagnose**: Audit tool logs and telemetry. Note if due to setup/assert.
+    - **Tip**: Proactively add custom logging/diagnostics to check hypotheses.
+
+---
+
+## 2. 🛠️ Fix Strategy
+
+1.  **Targeted Location**: Locate the test case and the corresponding
+    prompt/code.
+2.  **Iterative Scope**: Make extreme change first to verify scope, then refine
+    to a minimal, targeted change.
+3.  **Assertion Fidelity**:
+    - Changing the test prompt is a **last resort** (prompts are often vague by
+      design).
+    - **Warning**: Do not lose test fidelity by making prompts too direct/easy.
+    - **Primary Fix Trigger**: Adjust tool descriptions, system prompts
+      (`snippets.ts`), or **modules that contribute to the prompt template**.
+    - **Warning**: Prompts have multiple configurations; ensure your fix targets
+      the correct config for the model in question.
+4.  **Architecture Options**: If prompt or instruction tuning triggers no
+    improvement, analyze loop composition.
+    - **AgentLoop**: Defined by `context + toolset + prompt`.
+    - **Enhancements**: Loops perform best with direct prompts, fewer irrelevant
+      tools, low goal density, and minimal low-value/irrelevant context.
+    - **Modifications**: Compose subagents or isolate tools. Ground in observed
+      traces.
+    - **Warning**: Think deeply before offering recommendations; avoid parroting
+      abstract design guidelines.
+
+---
+
+## 3. ✅ Verify
+
+1.  **Run Local**: Run Vitest in non-interactive mode on just the file.
+2.  **Log Audit**: Prioritize diagnosing failures via log comparison before
+    triggering heavy test runs.
+3.  **Stability Limit**: Run the test **3 times** locally on key models (can use
+    scripts to run in parallel for speed):
+    - **Gemini 3.0**
+    - **Gemini 3 Flash**
+    - **Gemini 2.5 Pro**
+4.  **Flakiness Rule**: If it passes 2/3 times, it may be inherent noise
+    difficult to improve without a structural split.
+
+---
+
+## 4. 📊 Report
+
+Provide a summary of:
+
+- Test success rate for each tested model (e.g., 3/3 = 100%).
+- Root cause identification and fix explanation.
+- If unfixed, provide high-confidence architecture recommendations.

.gemini/skills/behavioral-evals/references/promoting.md

@@ -0,0 +1,55 @@
+# Promoting Behavioral Evals
+
+Use this guide when asked to analyze nightly results and promote incubated tests
+to stable suites.
+
+---
+
+## 1. 🔍 Investigate candidates
+
+1.  **Audit Nightly Logs**: Use the `gh` CLI to fetch results from
+    `evals-nightly.yml` (Direct URL:
+    `https://github.com/google-gemini/gemini-cli/actions/workflows/evals-nightly.yml`).
+    - **Tip**: The aggregate summary from the most recent run integrates the
+      last 7 runs of history automatically.
+    - **Safety**: DO NOT push changes or start remote runs. All verification is
+      local.
+2.  **Assess Stability**: Identify tests that pass **100% of the time** across
+    ALL enabled models over the **last 7 nightly runs** in a row.
+    - _100% means the test passed 3/3 times for every model and run._
+3.  **Promotion Targets**: Tests meeting this criteria are candidates for
+    promotion from `USUALLY_PASSES` to `ALWAYS_PASSES`.
+
+---
+
+## 2. 🚥 Promotion Steps
+
+1.  **Locate File**: Locate the eval file in the `evals/` directory.
+2.  **Update Policy**: Modify the policy argument to `ALWAYS_PASSES`.
+    ```typescript
+    evalTest('ALWAYS_PASSES', { ... })
+    ```
+3.  **Targeting**: Follow guidelines in `evals/README.md` regarding stable suite
+    organization.
+4.  **Constraint**: Your final change must be **minimal and targeted** strictly
+    to promoting the test status. Do not refactor the test or setup fixtures.
+
+---
+
+## 3. ✅ Verify
+
+1.  **Run Prompted Tests**: Run the promoted test locally using non-interactive
+    Vitest to confirm structure validity.
+2.  **Verify Suite Inclusion**: Check that the test is successfully picked up by
+    standard runnable ranges.
+
+---
+
+## 4. 📊 Report
+
+Provide a summary of:
+
+- Which tests were promoted.
+- Provide the success rate evidence (e.g., 7/7 runs passed for all models).
+- If no candidates qualified, list the next closest candidates and their current
+  pass rate.

.gemini/skills/behavioral-evals/references/running.md

@@ -0,0 +1,95 @@
+# Running & Promoting Evals
+
+## 🛠️ Prerequisites
+
+Behavioral evals run against the compiled binary. You **must** build and bundle
+the project first after making changes:
+
+```bash
+npm run build && npm run bundle
+```
+
+---
+
+## 🏃‍♂️ Running Tests
+
+### 1. Configure Environment Variables
+
+Evals require a standard API key. If your `.env` file has multiple keys or
+comments, use this precise extraction setup:
+
+```bash
+export GEMINI_API_KEY=$(grep '^GEMINI_API_KEY=' .env | cut -d '=' -f2) && RUN_EVALS=1 npx vitest run --config evals/vitest.config.ts <file_name>
+```
+
+### 2. Commands
+
+| Command                             | Scope           | Description                                        |
+| :---------------------------------- | :-------------- | :------------------------------------------------- |
+| `npm run test:always_passing_evals` | `ALWAYS_PASSES` | Fast feedback, runs in CI.                         |
+| `npm run test:all_evals`            | All             | Runs nightly incubation tests. Sets `RUN_EVALS=1`. |
+
+### Target Specific File
+
+_Note: `RUN_EVALS=1` is required for incubated (`USUALLY_PASSES`) tests._
+
+```bash
+RUN_EVALS=1 npx vitest run --config evals/vitest.config.ts my_feature.eval.ts
+```
+
+---
+
+## 🐞 Debugging and Logs
+
+If a test fails, verify:
+
+- **Tool Trajectory Logs**:序列 of calls in `evals/logs/<test_name>.log`.
+- **Verbose Reasoning**: Capture raw buffer traces by setting
+  `GEMINI_DEBUG_LOG_FILE`:
+  ```bash
+  export GEMINI_DEBUG_LOG_FILE="debug.log"
+  ```
+
+---
+
+### 🎯 Verify Model Targeting
+
+- **Tip:** Standard evals benchmark against model variations. If a test passes
+  on Flash but fails on Pro (or vice versa), the issue is usually in the **tool
+  description**, not the prompt definition. Flash is sensitive to "instruction
+  bloat," while Pro is sensitive to "ambiguous intent."
+
+---
+
+## 🚥 deflaking & Promotion
+
+To maintain CI stability, all new evals follow a strict incubation period.
+
+### 1. Incubation (`USUALLY_PASSES`)
+
+New tests must be created with the `USUALLY_PASSES` policy.
+
+```typescript
+evalTest('USUALLY_PASSES', { ... })
+```
+
+They run in **Evals: Nightly** workflows and do not block PR merges.
+
+### 2. Investigate Failures
+
+If a nightly eval regresses, investigate via agent:
+
+```bash
+gemini /fix-behavioral-eval [optional-run-uri]
+```
+
+### 3. Promotion (`ALWAYS_PASSES`)
+
+Once a test scores 100% consistency over multiple nightly cycles:
+
+```bash
+gemini /promote-behavioral-eval
+```
+
+_Do not promote manually._ The command verifies trajectory logs before updating
+the file policy.

.gemini/skills/ci/SKILL.md

@@ -0,0 +1,66 @@
+---
+name: ci
+description:
+  A specialized skill for Gemini CLI that provides high-performance, fail-fast
+  monitoring of GitHub Actions workflows and automated local verification of CI
+  failures. It handles run discovery automatically—simply provide the branch name.
+---
+
+# CI Replicate & Status
+
+This skill enables the agent to efficiently monitor GitHub Actions, triage
+failures, and bridge remote CI errors to local development. It defaults to
+**automatic replication** of failures to streamline the fix cycle.
+
+## Core Capabilities
+
+- **Automatic Replication**: Automatically monitors CI and immediately executes 
+  suggested test or lint commands locally upon failure.
+- **Real-time Monitoring**: Aggregated status line for all concurrent workflows
+  on the current branch.
+- **Fail-Fast Triage**: Immediately stops on the first job failure to provide a
+  structured report.
+
+## Workflow
+
+### 1. CI Replicate (`replicate`) - DEFAULT
+Use this as the primary path to monitor CI and **automatically** replicate 
+failures locally for immediate triage and fixing.
+- **Behavior**: When this workflow is triggered, the agent will monitor the CI
+  and **immediately and automatically execute** all suggested test or lint
+  commands (marked with 🚀) as soon as a failure is detected. 
+- **Tool**: `node .gemini/skills/ci/scripts/ci.mjs [branch]`
+- **Discovery**: The script **automatically** finds the latest active or recent
+  run for the branch. Do NOT manually search for run IDs.
+- **Goal**: Reproduce the failure locally without manual intervention, then
+  proceed to analyze and fix the code.
+
+### 1. CI Status (`status`)
+Use this when you have pushed changes and need to monitor the CI and reproduce
+any failures locally.
+- **Tool**: `node .gemini/skills/ci/scripts/ci.mjs [branch] [run_id]`
+- **Discovery**: The script **automatically** finds the latest active or recent
+  run for the branch. You should NOT manually search for \`run_id\` using \`gh run list\`
+  unless a specific historical run is requested. Simply provide the branch name.
+- **Step 1 (Monitor)**: Execute the tool with the branch name.
+- **Step 2 (Extract)**: Extract suggested \`npm test\` or \`npm run lint\` commands
+  from the output (marked with 🚀).
+- **Step 3 (Reproduce)**: Execute those commands locally to confirm the failure.
+- **Behavior**: It will poll every 15 seconds. If it detects a failure, it will
+  exit with a structured report and provide the exact commands to run locally.
+
+## Failure Categories & Actions
+
+- **Test Failures**: Agent should run the specific `npm test -w <pkg> -- <path>`
+  command suggested.
+- **Lint Errors**: Agent should run `npm run lint:all` or the specific package
+  lint command.
+- **Build Errors**: Agent should check `tsc` output or build logs to resolve
+  compilation issues.
+- **Job Errors**: Investigate `gh run view --job <job_id> --log` for
+  infrastructure or setup failures.
+
+## Noise Filtering
+The underlying scripts automatically filter noise (Git logs, NPM warnings, stack
+trace overhead). The agent should focus on the "Structured Failure Report"
+provided by the tool.

.gemini/skills/ci/scripts/ci.mjs

@@ -0,0 +1,281 @@
+#!/usr/bin/env node
+
+/**
+ * @license
+ * Copyright 2026 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { execSync } from 'node:child_process';
+
+const BRANCH =
+  process.argv[2] || execSync('git branch --show-current').toString().trim();
+const RUN_ID_OVERRIDE = process.argv[3];
+
+let REPO;
+try {
+  const remoteUrl = execSync('git remote get-url origin').toString().trim();
+  REPO = remoteUrl
+    .replace(/.*github\.com[\/:]/, '')
+    .replace(/\.git$/, '')
+    .trim();
+} catch (e) {
+  REPO = 'google-gemini/gemini-cli';
+}
+
+const FAILED_FILES = new Set();
+
+function runGh(args) {
+  try {
+    return execSync(`gh ${args}`, {
+      stdio: ['ignore', 'pipe', 'ignore'],
+    }).toString();
+  } catch (e) {
+    return null;
+  }
+}
+
+function fetchFailuresViaApi(jobId) {
+  try {
+    const cmd = `gh api repos/${REPO}/actions/jobs/${jobId}/logs | grep -iE " FAIL |❌|ERROR|Lint failed|Build failed|Exception|failed with exit code"`;
+    return execSync(cmd, {
+      stdio: ['ignore', 'pipe', 'ignore'],
+      maxBuffer: 10 * 1024 * 1024,
+    }).toString();
+  } catch (e) {
+    return '';
+  }
+}
+
+function isNoise(line) {
+  const lower = line.toLowerCase();
+  return (
+    lower.includes('* [new branch]') ||
+    lower.includes('npm warn') ||
+    lower.includes('fetching updates') ||
+    lower.includes('node:internal/errors') ||
+    lower.includes('at ') || // Stack traces
+    lower.includes('checkexecsyncerror') ||
+    lower.includes('node_modules')
+  );
+}
+
+function extractTestFile(failureText) {
+  const cleanLine = failureText
+    .replace(/[|#\[\]()]/g, ' ')
+    .replace(/<[^>]*>/g, ' ')
+    .trim();
+  const fileMatch = cleanLine.match(/([\w\/._-]+\.test\.[jt]sx?)/);
+  if (fileMatch) return fileMatch[1];
+  return null;
+}
+
+function generateTestCommand(failedFilesMap) {
+  const workspaceToFiles = new Map();
+  for (const [file, info] of failedFilesMap.entries()) {
+    if (
+      ['Job Error', 'Unknown File', 'Build Error', 'Lint Error'].includes(file)
+    )
+      continue;
+    let workspace = '@google/gemini-cli';
+    let relPath = file;
+    if (file.startsWith('packages/core/')) {
+      workspace = '@google/gemini-cli-core';
+      relPath = file.replace('packages/core/', '');
+    } else if (file.startsWith('packages/cli/')) {
+      workspace = '@google/gemini-cli';
+      relPath = file.replace('packages/cli/', '');
+    }
+    relPath = relPath.replace(/^.*packages\/[^\/]+\//, '');
+    if (!workspaceToFiles.has(workspace))
+      workspaceToFiles.set(workspace, new Set());
+    workspaceToFiles.get(workspace).add(relPath);
+  }
+  const commands = [];
+  for (const [workspace, files] of workspaceToFiles.entries()) {
+    commands.push(`npm test -w ${workspace} -- ${Array.from(files).join(' ')}`);
+  }
+  return commands.join(' && ');
+}
+
+async function monitor() {
+  let targetRunIds = [];
+  if (RUN_ID_OVERRIDE) {
+    targetRunIds = [RUN_ID_OVERRIDE];
+  } else {
+    // 1. Get runs directly associated with the branch
+    const runListOutput = runGh(
+      `run list --branch "${BRANCH}" --limit 10 --json databaseId,status,workflowName,createdAt`,
+    );
+    if (runListOutput) {
+      const runs = JSON.parse(runListOutput);
+      const activeRuns = runs.filter((r) => r.status !== 'completed');
+      if (activeRuns.length > 0) {
+        targetRunIds = activeRuns.map((r) => r.databaseId);
+      } else if (runs.length > 0) {
+        const latestTime = new Date(runs[0].createdAt).getTime();
+        targetRunIds = runs
+          .filter((r) => latestTime - new Date(r.createdAt).getTime() < 60000)
+          .map((r) => r.databaseId);
+      }
+    }
+
+    // 2. Get runs associated with commit statuses (handles chained/indirect runs)
+    try {
+      const headSha = execSync(`git rev-parse "${BRANCH}"`).toString().trim();
+      const statusOutput = runGh(
+        `api repos/${REPO}/commits/${headSha}/status -q '.statuses[] | select(.target_url | contains("actions/runs/")) | .target_url'`,
+      );
+      if (statusOutput) {
+        const statusRunIds = statusOutput
+          .split('\n')
+          .filter(Boolean)
+          .map((url) => {
+            const match = url.match(/actions\/runs\/(\d+)/);
+            return match ? parseInt(match[1], 10) : null;
+          })
+          .filter(Boolean);
+
+        for (const runId of statusRunIds) {
+          if (!targetRunIds.includes(runId)) {
+            targetRunIds.push(runId);
+          }
+        }
+      }
+    } catch (e) {
+      // Ignore if branch/SHA not found or API fails
+    }
+
+    if (targetRunIds.length > 0) {
+      const runNames = [];
+      for (const runId of targetRunIds) {
+        const runInfo = runGh(`run view "${runId}" --json workflowName`);
+        if (runInfo) {
+          runNames.push(JSON.parse(runInfo).workflowName);
+        }
+      }
+      console.log(`Monitoring workflows: ${[...new Set(runNames)].join(', ')}`);
+    }
+  }
+
+  if (targetRunIds.length === 0) {
+    console.log(`No runs found for branch ${BRANCH}.`);
+    process.exit(0);
+  }
+
+  while (true) {
+    let allPassed = 0,
+      allFailed = 0,
+      allRunning = 0,
+      allQueued = 0,
+      totalJobs = 0;
+    let anyRunInProgress = false;
+    const fileToTests = new Map();
+    let failuresFoundInLoop = false;
+
+    for (const runId of targetRunIds) {
+      const runOutput = runGh(
+        `run view "${runId}" --json databaseId,status,conclusion,workflowName`,
+      );
+      if (!runOutput) continue;
+      const run = JSON.parse(runOutput);
+      if (run.status !== 'completed') anyRunInProgress = true;
+
+      const jobsOutput = runGh(`run view "${runId}" --json jobs`);
+      if (jobsOutput) {
+        const { jobs } = JSON.parse(jobsOutput);
+        totalJobs += jobs.length;
+        const failedJobs = jobs.filter((j) => j.conclusion === 'failure');
+        if (failedJobs.length > 0) {
+          failuresFoundInLoop = true;
+          for (const job of failedJobs) {
+            const failures = fetchFailuresViaApi(job.databaseId);
+            if (failures.trim()) {
+              failures.split('\n').forEach((line) => {
+                if (!line.trim() || isNoise(line)) return;
+                const file = extractTestFile(line);
+                const filePath =
+                  file ||
+                  (line.toLowerCase().includes('lint')
+                    ? 'Lint Error'
+                    : line.toLowerCase().includes('build')
+                      ? 'Build Error'
+                      : 'Unknown File');
+                let testName = line;
+                if (line.includes(' > ')) {
+                  testName = line.split(' > ').slice(1).join(' > ').trim();
+                }
+                if (!fileToTests.has(filePath))
+                  fileToTests.set(filePath, new Set());
+                fileToTests.get(filePath).add(testName);
+              });
+            } else {
+              const step =
+                job.steps?.find((s) => s.conclusion === 'failure')?.name ||
+                'unknown';
+              const category = step.toLowerCase().includes('lint')
+                ? 'Lint Error'
+                : step.toLowerCase().includes('build')
+                  ? 'Build Error'
+                  : 'Job Error';
+              if (!fileToTests.has(category))
+                fileToTests.set(category, new Set());
+              fileToTests
+                .get(category)
+                .add(`${job.name}: Failed at step "${step}"`);
+            }
+          }
+        }
+        for (const job of jobs) {
+          if (job.status === 'in_progress') allRunning++;
+          else if (job.status === 'queued') allQueued++;
+          else if (job.conclusion === 'success') allPassed++;
+          else if (job.conclusion === 'failure') allFailed++;
+        }
+      }
+    }
+
+    if (failuresFoundInLoop) {
+      console.log(
+        `\n\n❌ Failures detected across ${allFailed} job(s). Stopping monitor...`,
+      );
+      console.log('\n--- Structured Failure Report (Noise Filtered) ---');
+      for (const [file, tests] of fileToTests.entries()) {
+        console.log(`\nCategory/File: ${file}`);
+        // Limit output per file if it's too large
+        const testsArr = Array.from(tests).map((t) =>
+          t.length > 500 ? t.substring(0, 500) + '... [TRUNCATED]' : t,
+        );
+        testsArr.slice(0, 10).forEach((t) => console.log(`  - ${t}`));
+        if (testsArr.length > 10)
+          console.log(`  ... and ${testsArr.length - 10} more`);
+      }
+      const testCmd = generateTestCommand(fileToTests);
+      if (testCmd) {
+        console.log('\n🚀 Run this to verify fixes:');
+        console.log(testCmd);
+      } else if (
+        Array.from(fileToTests.keys()).some((k) => k.includes('Lint'))
+      ) {
+        console.log('\n🚀 Run this to verify lint fixes:\nnpm run lint:all');
+      }
+      console.log('---------------------------------');
+      process.exit(1);
+    }
+
+    const completed = allPassed + allFailed;
+    process.stdout.write(
+      `\r⏳ Monitoring ${targetRunIds.length} runs... ${completed}/${totalJobs} jobs (${allPassed} passed, ${allFailed} failed, ${allRunning} running, ${allQueued} queued)          `,
+    );
+    if (!anyRunInProgress) {
+      console.log('\n✅ All workflows passed!');
+      process.exit(0);
+    }
+    await new Promise((r) => setTimeout(r, 15000));
+  }
+}
+
+monitor().catch((err) => {
+  console.error('\nMonitor error:', err.message);
+  process.exit(1);
+});

.gemini/skills/docs-writer/SKILL.md

@@ -71,12 +71,44 @@ accessible.
   tables).
 - **Media:** Use lowercase hyphenated filenames. Provide descriptive alt text
   for all images.
+- **Details section:** Use the `<details>` tag to create a collapsible section.
+  This is useful for supplementary or data-heavy information that isn't critical
+  to the main flow.
+
+  Example:
+
+  <details>
+  <summary>Title</summary>
+
+  - First entry
+  - Second entry
+
+  </details>
+
+- **Callouts**: Use GitHub-flavored markdown alerts to highlight important
+  information. To ensure the formatting is preserved by `npm run format`, place
+  an empty line, then the `<!-- prettier-ignore -->` comment directly before
+  the callout block. The callout type (`[!TYPE]`) should be on the first line,
+  followed by a newline, and then the content, with each subsequent line of
+  content starting with `>`. Available types are `NOTE`, `TIP`, `IMPORTANT`,
+  `WARNING`, and `CAUTION`.
+
+  Example:
+
+<!-- prettier-ignore -->
+> [!NOTE]
+> This is an example of a multi-line note that will be preserved
+> by Prettier.
 
 ### Structure
 - **BLUF:** Start with an introduction explaining what to expect.
 - **Experimental features:** If a feature is clearly noted as experimental,
-add the following note immediately after the introductory paragraph:
-  `> **Note:** This is a preview feature currently under active development.`
+  add the following note immediately after the introductory paragraph:
+
+<!-- prettier-ignore -->
+> [!NOTE]
+> This is an experimental feature currently under active development.
+
 - **Headings:** Use hierarchical headings to support the user journey.
 - **Procedures:** 
   - Introduce lists of steps with a complete sentence.
@@ -85,8 +117,7 @@ add the following note immediately after the introductory paragraph:
   - Put conditions before instructions (e.g., "On the Settings page, click...").
   - Provide clear context for where the action takes place.
   - Indicate optional steps clearly (e.g., "Optional: ...").
-- **Elements:** Use bullet lists, tables, notes (`> **Note:**`), and warnings 
-  (`> **Warning:**`).
+- **Elements:** Use bullet lists, tables, details, and callouts.
 - **Avoid using a table of contents:** If a table of contents is present, remove
   it.
 - **Next steps:** Conclude with a "Next steps" section if applicable.

.geminiignore

@@ -0,0 +1 @@
+packages/core/src/services/scripts/*.exe

.github/ISSUE_TEMPLATE/website_issue.yml

@@ -1,7 +1,9 @@
 name: 'Website issue'
 description: 'Report an issue with the Gemini CLI Website and Gemini CLI Extensions Gallery'
+title: 'GeminiCLI.com Feedback: [ISSUE]'
 labels:
   - 'area/extensions'
+  - 'area/documentation'
 body:
   - type: 'markdown'
     attributes:

.github/workflows/eval-guidance.yml

@@ -0,0 +1,69 @@
+name: 'Evals: PR Guidance'
+
+on:
+  pull_request:
+    paths:
+      - 'packages/core/src/**/*.ts'
+      - '!**/*.test.ts'
+      - '!**/*.test.tsx'
+
+permissions:
+  pull-requests: 'write'
+  contents: 'read'
+
+jobs:
+  provide-guidance:
+    name: 'Model Steering Guidance'
+    runs-on: 'ubuntu-latest'
+    if: "github.repository == 'google-gemini/gemini-cli'"
+    steps:
+      - name: 'Checkout'
+        uses: 'actions/checkout@08eba0b27e820071cde6df949e0beb9ba4906955' # ratchet:actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: 'Set up Node.js'
+        uses: 'actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020' # ratchet:actions/setup-node@v4.4.0
+        with:
+          node-version-file: '.nvmrc'
+          cache: 'npm'
+
+      - name: 'Detect Steering Changes'
+        id: 'detect'
+        run: |
+          STEERING_DETECTED=$(node scripts/changed_prompt.js --steering-only)
+          echo "STEERING_DETECTED=$STEERING_DETECTED" >> "$GITHUB_OUTPUT"
+
+      - name: 'Analyze PR Content'
+        if: "steps.detect.outputs.STEERING_DETECTED == 'true'"
+        id: 'analysis'
+        env:
+          GH_TOKEN: '${{ secrets.GITHUB_TOKEN }}'
+        run: |
+          # Check for behavioral eval changes
+          EVAL_CHANGES=$(git diff --name-only origin/${{ github.base_ref }}...HEAD | grep "^evals/" || true)
+          if [ -z "$EVAL_CHANGES" ]; then
+            echo "MISSING_EVALS=true" >> "$GITHUB_OUTPUT"
+          fi
+
+          # Check if user is a maintainer (has write/admin access)
+          USER_PERMISSION=$(gh api repos/${{ github.repository }}/collaborators/${{ github.actor }}/permission --jq '.permission')
+          if [[ "$USER_PERMISSION" == "admin" || "$USER_PERMISSION" == "write" ]]; then
+            echo "IS_MAINTAINER=true" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: 'Post Guidance Comment'
+        if: "steps.detect.outputs.STEERING_DETECTED == 'true'"
+        uses: 'thollander/actions-comment-pull-request@65f9e5c9a1f2cd378bd74b2e057c9736982a8e74' # ratchet:thollander/actions-comment-pull-request@v3
+        with:
+          comment-tag: 'eval-guidance-bot'
+          message: |
+            ### 🧠 Model Steering Guidance
+
+            This PR modifies files that affect the model's behavior (prompts, tools, or instructions).
+
+            ${{ steps.analysis.outputs.MISSING_EVALS == 'true' && '- ⚠️ **Consider adding Evals:** No behavioral evaluations (`evals/*.eval.ts`) were added or updated in this PR. Consider adding a test case to verify the new behavior and prevent regressions.' || '' }}
+            ${{ steps.analysis.outputs.IS_MAINTAINER == 'true' && '- 🚀 **Maintainer Reminder:** Please ensure that these changes do not regress results on benchmark evals before merging.' || '' }}
+
+            ---
+            *This is an automated guidance message triggered by steering logic signatures.*

.github/workflows/evals-nightly.yml

@@ -61,6 +61,7 @@ jobs:
           GEMINI_MODEL: '${{ matrix.model }}'
           RUN_EVALS: "${{ github.event.inputs.run_all != 'false' }}"
           TEST_NAME_PATTERN: '${{ github.event.inputs.test_name_pattern }}'
+          VITEST_RETRY: 0
         run: |
           CMD="npm run test:all_evals"
           PATTERN="${TEST_NAME_PATTERN}"

docs/admin/enterprise-controls.md

@@ -106,6 +106,67 @@ organization.
   ensures users maintain final control over which permitted servers are actually
   active in their environment.
 
+#### Required MCP Servers (preview)
+
+**Default**: empty
+
+Allows administrators to define MCP servers that are **always injected** into
+the user's environment. Unlike the allowlist (which filters user-configured
+servers), required servers are automatically added regardless of the user's
+local configuration.
+
+**Required Servers Format:**
+
+```json
+{
+  "requiredMcpServers": {
+    "corp-compliance-tool": {
+      "url": "https://mcp.corp/compliance",
+      "type": "http",
+      "trust": true,
+      "description": "Corporate compliance tool"
+    },
+    "internal-registry": {
+      "url": "https://registry.corp/mcp",
+      "type": "sse",
+      "authProviderType": "google_credentials",
+      "oauth": {
+        "scopes": ["https://www.googleapis.com/auth/scope"]
+      }
+    }
+  }
+}
+```
+
+**Supported Fields:**
+
+- `url`: (Required) The full URL of the MCP server endpoint.
+- `type`: (Required) The connection type (`sse` or `http`).
+- `trust`: (Optional) If set to `true`, tool execution will not require user
+  approval. Defaults to `true` for required servers.
+- `description`: (Optional) Human-readable description of the server.
+- `authProviderType`: (Optional) Authentication provider (`dynamic_discovery`,
+  `google_credentials`, or `service_account_impersonation`).
+- `oauth`: (Optional) OAuth configuration including `scopes`, `clientId`, and
+  `clientSecret`.
+- `targetAudience`: (Optional) OAuth target audience for service-to-service
+  auth.
+- `targetServiceAccount`: (Optional) Service account email to impersonate.
+- `headers`: (Optional) Additional HTTP headers to send with requests.
+- `includeTools` / `excludeTools`: (Optional) Tool filtering lists.
+- `timeout`: (Optional) Timeout in milliseconds for MCP requests.
+
+**Client Enforcement Logic:**
+
+- Required servers are injected **after** allowlist filtering, so they are
+  always available even if the allowlist is active.
+- If a required server has the **same name** as a locally configured server, the
+  admin configuration **completely overrides** the local one.
+- Required servers only support remote transports (`sse`, `http`). Local
+  execution fields (`command`, `args`, `env`, `cwd`) are not supported.
+- Required servers can coexist with allowlisted servers — both features work
+  independently.
+
 ### Unmanaged Capabilities
 
 **Enabled/Disabled** | Default: disabled

docs/changelogs/preview.md

@@ -1,6 +1,6 @@
-# Preview release: v0.35.0-preview.1
+# Preview release: v0.35.0-preview.5
 
-Released: March 17, 2026
+Released: March 23, 2026
 
 Our preview release includes the latest, new, and experimental features. This
 release may not be as stable as our [latest weekly release](latest.md).
@@ -33,6 +33,17 @@ npm install -g @google/gemini-cli@preview
 
 ## What's Changed
 
+- fix(patch): cherry-pick b2d6dc4 to release/v0.35.0-preview.4-pr-23546
+  [CONFLICTS] by @gemini-cli-robot in
+  [#23585](https://github.com/google-gemini/gemini-cli/pull/23585)
+- fix(patch): cherry-pick daf3691 to release/v0.35.0-preview.2-pr-23558 to patch
+  version v0.35.0-preview.2 and create version 0.35.0-preview.3 by
+  @gemini-cli-robot in
+  [#23565](https://github.com/google-gemini/gemini-cli/pull/23565)
+- fix(patch): cherry-pick 4e5dfd0 to release/v0.35.0-preview.1-pr-23074 to patch
+  version v0.35.0-preview.1 and create version 0.35.0-preview.2 by
+  @gemini-cli-robot in
+  [#23134](https://github.com/google-gemini/gemini-cli/pull/23134)
 - feat(cli): customizable keyboard shortcuts by @scidomino in
   [#21945](https://github.com/google-gemini/gemini-cli/pull/21945)
 - feat(core): Thread `AgentLoopContext` through core. by @joshualitt in
@@ -373,4 +384,4 @@ npm install -g @google/gemini-cli@preview
   [#22815](https://github.com/google-gemini/gemini-cli/pull/22815)
 
 **Full Changelog**:
-https://github.com/google-gemini/gemini-cli/compare/v0.34.0-preview.4...v0.35.0-preview.1
+https://github.com/google-gemini/gemini-cli/compare/v0.34.0-preview.4...v0.35.0-preview.5

docs/cli/checkpointing.md

@@ -39,7 +39,9 @@ file in your project's temporary directory, typically located at
 The Checkpointing feature is disabled by default. To enable it, you need to edit
 your `settings.json` file.
 
-> **Note:** The `--checkpointing` command-line flag was removed in version
+<!-- prettier-ignore -->
+> [!CAUTION]
+> The `--checkpointing` command-line flag was removed in version
 > 0.11.0. Checkpointing can now only be enabled through the `settings.json`
 > configuration file.
 

docs/cli/cli-reference.md

@@ -50,6 +50,7 @@ These commands are available within the interactive REPL.
 | `--model`                        | `-m`  | string  | `auto`    | Model to use. See [Model Selection](#model-selection) for available values.                                                                                            |
 | `--prompt`                       | `-p`  | string  | -         | Prompt text. Appended to stdin input if provided. Forces non-interactive mode.                                                                                         |
 | `--prompt-interactive`           | `-i`  | string  | -         | Execute prompt and continue in interactive mode                                                                                                                        |
+| `--worktree`                     | `-w`  | string  | -         | Start Gemini in a new git worktree. If no name is provided, one is generated automatically. Requires `experimental.worktrees: true` in settings.                       |
 | `--sandbox`                      | `-s`  | boolean | `false`   | Run in a sandboxed environment for safer execution                                                                                                                     |
 | `--approval-mode`                | -     | string  | `default` | Approval mode for tool execution. Choices: `default`, `auto_edit`, `yolo`                                                                                              |
 | `--yolo`                         | `-y`  | boolean | `false`   | **Deprecated.** Auto-approve all actions. Use `--approval-mode=yolo` instead.                                                                                          |

docs/cli/custom-commands.md

@@ -30,7 +30,9 @@ separator (`/` or `\`) being converted to a colon (`:`).
 - A file at `<project>/.gemini/commands/git/commit.toml` becomes the namespaced
   command `/git:commit`.
 
-> [!TIP] After creating or modifying `.toml` command files, run
+<!-- prettier-ignore -->
+> [!TIP]
+> After creating or modifying `.toml` command files, run
 > `/commands reload` to pick up your changes without restarting the CLI.
 
 ## TOML file format (v1)
@@ -177,10 +179,10 @@ ensure that only intended commands can be run.
     automatically shell-escaped (see
     [Context-Aware Injection](#1-context-aware-injection-with-args) above).
 3.  **Robust parsing:** The parser correctly handles complex shell commands that
-    include nested braces, such as JSON payloads. **Note:** The content inside
-    `!{...}` must have balanced braces (`{` and `}`). If you need to execute a
-    command containing unbalanced braces, consider wrapping it in an external
-    script file and calling the script within the `!{...}` block.
+    include nested braces, such as JSON payloads. The content inside `!{...}`
+    must have balanced braces (`{` and `}`). If you need to execute a command
+    containing unbalanced braces, consider wrapping it in an external script
+    file and calling the script within the `!{...}` block.
 4.  **Security check and confirmation:** The CLI performs a security check on
     the final, resolved command (after arguments are escaped and substituted). A
     dialog will appear showing the exact command(s) to be executed.

docs/cli/enterprise.md

@@ -5,9 +5,11 @@ and managing Gemini CLI in an enterprise environment. By leveraging system-level
 settings, administrators can enforce security policies, manage tool access, and
 ensure a consistent experience for all users.
 
-> **A note on security:** The patterns described in this document are intended
-> to help administrators create a more controlled and secure environment for
-> using Gemini CLI. However, they should not be considered a foolproof security
+<!-- prettier-ignore -->
+> [!WARNING]
+> The patterns described in this document are intended to help
+> administrators create a more controlled and secure environment for using
+> Gemini CLI. However, they should not be considered a foolproof security
 > boundary. A determined user with sufficient privileges on their local machine
 > may still be able to circumvent these configurations. These measures are
 > designed to prevent accidental misuse and enforce corporate policy in a
@@ -280,10 +282,12 @@ environment to a blocklist.
 }
 ```
 
-**Security note:** Blocklisting with `excludeTools` is less secure than
-allowlisting with `coreTools`, as it relies on blocking known-bad commands, and
-clever users may find ways to bypass simple string-based blocks. **Allowlisting
-is the recommended approach.**
+<!-- prettier-ignore -->
+> [!WARNING]
+> Blocklisting with `excludeTools` is less secure than
+> allowlisting with `coreTools`, as it relies on blocking known-bad commands,
+> and clever users may find ways to bypass simple string-based blocks.
+> **Allowlisting is the recommended approach.**
 
 ### Disabling YOLO mode
 
@@ -494,8 +498,10 @@ other events. For more information, see the
 }
 ```
 
-**Note:** Ensure that `logPrompts` is set to `false` in an enterprise setting to
-avoid collecting potentially sensitive information from user prompts.
+<!-- prettier-ignore -->
+> [!NOTE]
+> Ensure that `logPrompts` is set to `false` in an enterprise setting to
+> avoid collecting potentially sensitive information from user prompts.
 
 ## Authentication
 

docs/cli/git-worktrees.md

@@ -0,0 +1,107 @@
+# Git Worktrees (experimental)
+
+When working on multiple tasks at once, you can use Git worktrees to give each
+Gemini session its own copy of the codebase. Git worktrees create separate
+working directories that each have their own files and branch while sharing the
+same repository history. This prevents changes in one session from colliding
+with another.
+
+Learn more about [session management](./session-management.md).
+
+<!-- prettier-ignore -->
+> [!NOTE]
+> This is an experimental feature currently under active development. Your
+> feedback is invaluable as we refine this feature. If you have ideas,
+> suggestions, or encounter issues:
+>
+> - [Open an issue](https://github.com/google-gemini/gemini-cli/issues/new?template=bug_report.yml) on GitHub.
+> - Use the **/bug** command within Gemini CLI to file an issue.
+
+Learn more in the official Git worktree
+[documentation](https://git-scm.com/docs/git-worktree).
+
+## How to enable Git worktrees
+
+Git worktrees are an experimental feature. You must enable them in your settings
+using the `/settings` command or by manually editing your `settings.json` file.
+
+1.  Use the `/settings` command.
+2.  Search for and set **Enable Git Worktrees** to `true`.
+
+Alternatively, add the following to your `settings.json`:
+
+```json
+{
+  "experimental": {
+    "worktrees": true
+  }
+}
+```
+
+## How to use Git worktrees
+
+Use the `--worktree` (`-w`) flag to create an isolated worktree and start Gemini
+CLI in it.
+
+- **Start with a specific name:** The value you pass becomes both the directory
+  name (within `.gemini/worktrees/`) and the branch name.
+
+  ```bash
+  gemini --worktree feature-search
+  ```
+
+- **Start with a random name:** If you omit the name, Gemini generates a random
+  one automatically (for example, `worktree-a1b2c3d4`).
+
+  ```bash
+  gemini --worktree
+  ```
+
+<!-- prettier-ignore -->
+> [!NOTE]
+> Remember to initialize your development environment in each new
+> worktree according to your project's setup. Depending on your stack, this
+> might include running dependency installation (`npm install`, `yarn`), setting
+> up virtual environments, or following your project's standard build process.
+
+## How to exit a Git worktree session
+
+When you exit a worktree session (using `/quit` or `Ctrl+C`), Gemini leaves the
+worktree intact so your work is not lost. This includes your uncommitted changes
+(modified files, staged changes, or untracked files) and any new commits you
+have made.
+
+Gemini prioritizes a fast and safe exit: it **does not automatically delete**
+your worktree or branch. You are responsible for cleaning up your worktrees
+manually once you are finished with them.
+
+When you exit, Gemini displays instructions on how to resume your work or how to
+manually remove the worktree if you no longer need it.
+
+## Resuming work in a Git worktree
+
+To resume a session in a worktree, navigate to the worktree directory and start
+Gemini CLI with the `--resume` flag and the session ID:
+
+```bash
+cd .gemini/worktrees/feature-search
+gemini --resume <session_id>
+```
+
+## Managing Git worktrees manually
+
+For more control over worktree location and branch configuration, or to clean up
+a preserved worktree, you can use Git directly:
+
+- **Clean up a preserved Git worktree:**
+  ```bash
+  git worktree remove .gemini/worktrees/feature-search --force
+  git branch -D worktree-feature-search
+  ```
+- **Create a Git worktree manually:**
+  ```bash
+  git worktree add ../project-feature-search -b feature-search
+  cd ../project-feature-search && gemini
+  ```
+
+[Open an issue]: https://github.com/google-gemini/gemini-cli/issues

docs/cli/model-steering.md

@@ -4,9 +4,10 @@ Model steering lets you provide real-time guidance and feedback to Gemini CLI
 while it is actively executing a task. This lets you correct course, add missing
 context, or skip unnecessary steps without having to stop and restart the agent.
 
-> **Note:** This is a preview feature under active development. Preview features
-> may only be available in the **Preview** channel or may need to be enabled
-> under `/settings`.
+<!-- prettier-ignore -->
+> [!NOTE]
+> This is an experimental feature currently under active development and
+> may need to be enabled under `/settings`.
 
 Model steering is particularly useful during complex [Plan Mode](./plan-mode.md)
 workflows or long-running subagent executions where you want to ensure the agent

docs/cli/model.md

@@ -5,7 +5,9 @@ used by Gemini CLI, giving you more control over your results. Use **Pro**
 models for complex tasks and reasoning, **Flash** models for high speed results,
 or the (recommended) **Auto** setting to choose the best model for your tasks.
 
-> **Note:** The `/model` command (and the `--model` flag) does not override the
+<!-- prettier-ignore -->
+> [!NOTE]
+> The `/model` command (and the `--model` flag) does not override the
 > model used by sub-agents. Consequently, even when using the `/model` flag you
 > may see other models used in your model usage reports.
 

docs/cli/notifications.md

@@ -4,9 +4,10 @@ Gemini CLI can send system notifications to alert you when a session completes
 or when it needs your attention, such as when it's waiting for you to approve a
 tool call.
 
-> **Note:** This is a preview feature currently under active development.
-> Preview features may be available on the **Preview** channel or may need to be
-> enabled under `/settings`.
+<!-- prettier-ignore -->
+> [!NOTE]
+> This is an experimental feature currently under active development and
+> may need to be enabled under `/settings`.
 
 Notifications are particularly useful when running long-running tasks or using
 [Plan Mode](./plan-mode.md), letting you switch to other windows while Gemini

docs/cli/plan-mode.md

@@ -35,19 +35,17 @@ To launch Gemini CLI in Plan Mode once:
 To start Plan Mode while using Gemini CLI:
 
 - **Keyboard shortcut:** Press `Shift+Tab` to cycle through approval modes
-  (`Default` -> `Auto-Edit` -> `Plan`).
-
-  > **Note:** Plan Mode is automatically removed from the rotation when Gemini
-  > CLI is actively processing or showing confirmation dialogs.
+  (`Default` -> `Auto-Edit` -> `Plan`). Plan Mode is automatically removed from
+  the rotation when Gemini CLI is actively processing or showing confirmation
+  dialogs.
 
 - **Command:** Type `/plan` in the input box.
 
 - **Natural Language:** Ask Gemini CLI to "start a plan for...". Gemini CLI
   calls the
   [`enter_plan_mode`](../tools/planning.md#1-enter_plan_mode-enterplanmode) tool
-  to switch modes.
-  > **Note:** This tool is not available when Gemini CLI is in
-  > [YOLO mode](../reference/configuration.md#command-line-arguments).
+  to switch modes. This tool is not available when Gemini CLI is in
+  [YOLO mode](../reference/configuration.md#command-line-arguments).
 
 ## How to use Plan Mode
 
@@ -202,6 +200,7 @@ your specific environment.
 
 ```toml
 [[rule]]
+toolName = "*"
 mcpName = "*"
 toolAnnotations = { readOnlyHint = true }
 decision = "allow"
@@ -407,7 +406,9 @@ To build a custom planning workflow, you can use:
   [custom plan directories](#custom-plan-directory-and-policies) and
   [custom policies](#custom-policies).
 
-> **Note:** Use [Conductor] as a reference when building your own custom
+<!-- prettier-ignore -->
+> [!TIP]
+> Use [Conductor] as a reference when building your own custom
 > planning workflow.
 
 By using Plan Mode as its execution environment, your custom methodology can

docs/cli/sandbox.md

@@ -50,7 +50,25 @@ Cross-platform sandboxing with complete process isolation.
 **Note**: Requires building the sandbox image locally or using a published image
 from your organization's registry.
 
-### 3. gVisor / runsc (Linux only)
+### 3. Windows Native Sandbox (Windows only)
+
+... **Troubleshooting and Side Effects:**
+
+The Windows Native sandbox uses the `icacls` command to set a "Low Mandatory
+Level" on files and directories it needs to write to.
+
+- **Persistence**: These integrity level changes are persistent on the
+  filesystem. Even after the sandbox session ends, files created or modified by
+  the sandbox will retain their "Low" integrity level.
+- **Manual Reset**: If you need to reset the integrity level of a file or
+  directory, you can use:
+  ```powershell
+  icacls "C:\path\to\dir" /setintegritylevel Medium
+  ```
+- **System Folders**: The sandbox manager automatically skips setting integrity
+  levels on system folders (like `C:\Windows`) for safety.
+
+### 4. gVisor / runsc (Linux only)
 
 Strongest isolation available: runs containers inside a user-space kernel via
 [gVisor](https://github.com/google/gvisor). gVisor intercepts all container
@@ -253,9 +271,11 @@ $env:SANDBOX_SET_UID_GID="false"  # Disable UID/GID mapping
 DEBUG=1 gemini -s -p "debug command"
 ```
 
-**Note:** If you have `DEBUG=true` in a project's `.env` file, it won't affect
-gemini-cli due to automatic exclusion. Use `.gemini/.env` files for gemini-cli
-specific debug settings.
+<!-- prettier-ignore -->
+> [!NOTE]
+> If you have `DEBUG=true` in a project's `.env` file, it won't affect
+> gemini-cli due to automatic exclusion. Use `.gemini/.env` files for
+> gemini-cli specific debug settings.
 
 ### Inspect sandbox
 

docs/cli/session-management.md

@@ -96,6 +96,12 @@ Compatibility aliases:
 - `/chat ...` works for the same commands.
 - `/resume checkpoints ...` also remains supported during migration.
 
+## Parallel sessions with Git worktrees
+
+When working on multiple tasks at once, you can use
+[Git worktrees](./git-worktrees.md) to give each Gemini session its own copy of
+the codebase. This prevents changes in one session from colliding with another.
+
 ## Managing sessions
 
 You can list and delete sessions to keep your history organized and manage disk

docs/cli/settings.md

@@ -11,7 +11,9 @@ locations:
 - **User settings**: `~/.gemini/settings.json`
 - **Workspace settings**: `your-project/.gemini/settings.json`
 
-Note: Workspace settings override user settings.
+<!-- prettier-ignore -->
+> [!IMPORTANT]
+> Workspace settings override user settings.
 
 ## Settings reference
 
@@ -99,6 +101,13 @@ they appear in the UI.
 | Disable Loop Detection        | `model.disableLoopDetection` | Disable automatic detection and prevention of infinite loops.                          | `false`     |
 | Skip Next Speaker Check       | `model.skipNextSpeakerCheck` | Skip the next speaker check.                                                           | `true`      |
 
+### Agents
+
+| UI Label                  | Setting                                  | Description                                                                                   | Default |
+| ------------------------- | ---------------------------------------- | --------------------------------------------------------------------------------------------- | ------- |
+| Confirm Sensitive Actions | `agents.browser.confirmSensitiveActions` | Require manual confirmation for sensitive browser actions (e.g., fill_form, evaluate_script). | `false` |
+| Block File Uploads        | `agents.browser.blockFileUploads`        | Hard-block file upload requests from the browser agent.                                       | `false` |
+
 ### Context
 
 | UI Label                             | Setting                                           | Description                                                                                                                                                                                                                                 | Default |
@@ -115,6 +124,8 @@ they appear in the UI.
 
 | UI Label                         | Setting                              | Description                                                                                                                                                                | Default |
 | -------------------------------- | ------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------- |
+| Sandbox Allowed Paths            | `tools.sandboxAllowedPaths`          | List of additional paths that the sandbox is allowed to access.                                                                                                            | `[]`    |
+| Sandbox Network Access           | `tools.sandboxNetworkAccess`         | Whether the sandbox is allowed to access the network.                                                                                                                      | `false` |
 | Enable Interactive Shell         | `tools.shell.enableInteractiveShell` | Use node-pty for an interactive shell experience. Fallback to child_process still applies.                                                                                 | `true`  |
 | Show Color                       | `tools.shell.showColor`              | Show color in shell output.                                                                                                                                                | `false` |
 | Use Ripgrep                      | `tools.useRipgrep`                   | Use ripgrep for file content search instead of the fallback implementation. Provides faster search performance.                                                            | `true`  |
@@ -147,11 +158,13 @@ they appear in the UI.
 | UI Label                   | Setting                                  | Description                                                                                                                                               | Default |
 | -------------------------- | ---------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------- | ------- |
 | Enable Tool Output Masking | `experimental.toolOutputMasking.enabled` | Enables tool output masking to save tokens.                                                                                                               | `true`  |
+| Enable Git Worktrees       | `experimental.worktrees`                 | Enable automated Git worktree management for parallel work.                                                                                               | `false` |
 | Use OSC 52 Paste           | `experimental.useOSC52Paste`             | Use OSC 52 for pasting. This may be more robust than the default system when using remote terminal sessions (if your terminal is configured to allow it). | `false` |
 | Use OSC 52 Copy            | `experimental.useOSC52Copy`              | Use OSC 52 for copying. This may be more robust than the default system when using remote terminal sessions (if your terminal is configured to allow it). | `false` |
 | Plan                       | `experimental.plan`                      | Enable Plan Mode.                                                                                                                                         | `true`  |
 | Model Steering             | `experimental.modelSteering`             | Enable model steering (user hints) to guide the model during tool execution.                                                                              | `false` |
 | Direct Web Fetch           | `experimental.directWebFetch`            | Enable web fetch behavior that bypasses LLM summarization.                                                                                                | `false` |
+| Memory Manager Agent       | `experimental.memoryManager`             | Replace the built-in save_memory tool with a memory manager subagent that supports adding, removing, de-duplicating, and organizing memories.             | `false` |
 | Topic & Update Narration   | `experimental.topicUpdateNarration`      | Enable the experimental Topic & Update communication model for reduced chattiness and structured progress reporting.                                      | `false` |
 
 ### Skills

docs/cli/skills.md

@@ -63,8 +63,10 @@ Use the `/skills` slash command to view and manage available expertise:
 - `/skills enable <name>`: Re-enables a disabled skill.
 - `/skills reload`: Refreshes the list of discovered skills from all tiers.
 
-_Note: `/skills disable` and `/skills enable` default to the `user` scope. Use
-`--scope workspace` to manage workspace-specific settings._
+<!-- prettier-ignore -->
+> [!NOTE]
+> `/skills disable` and `/skills enable` default to the `user` scope. Use
+> `--scope workspace` to manage workspace-specific settings.
 
 ### From the Terminal
 

docs/cli/system-prompt.md

@@ -14,7 +14,9 @@ core instructions will apply unless you include them yourself.
 This feature is intended for advanced users who need to enforce strict,
 project-specific behavior or create a customized persona.
 
-> Tip: You can export the current default system prompt to a file first, review
+<!-- prettier-ignore -->
+> [!TIP]
+> You can export the current default system prompt to a file first, review
 > it, and then selectively modify or replace it (see
 > [“Export the default prompt”](#export-the-default-prompt-recommended)).
 

docs/cli/telemetry.md

@@ -125,9 +125,11 @@ You must complete several setup steps before enabling Google Cloud telemetry.
       }
       ```
 
-      > **Note:** This setting requires **Direct export** (in-process exporters)
-      > and cannot be used when `useCollector` is `true`. If both are enabled,
-      > telemetry will be disabled.
+<!-- prettier-ignore -->
+> [!NOTE]
+> This setting requires **Direct export** (in-process exporters)
+> and cannot be used when `useCollector` is `true`. If both are enabled,
+> telemetry will be disabled.
 
 3.  Ensure your account or service account has these IAM roles:
     - Cloud Trace Agent
@@ -304,6 +306,7 @@ Emitted at startup with the CLI configuration.
 - `extension_ids` (string)
 - `extensions_count` (int)
 - `auth_type` (string)
+- `worktree_active` (boolean)
 - `github_workflow_name` (string, optional)
 - `github_repository_hash` (string, optional)
 - `github_event_name` (string, optional)
@@ -901,6 +904,20 @@ Logs keychain availability checks.
 
 - `available` (boolean)
 
+##### `gemini_cli.startup_stats`
+
+Logs detailed startup performance statistics.
+
+<details>
+<summary>Attributes</summary>
+
+- `phases` (json array of startup phases)
+- `os_platform` (string)
+- `os_release` (string)
+- `is_docker` (boolean)
+
+</details>
+
 </details>
 
 ### Metrics
@@ -917,6 +934,20 @@ Gemini CLI exports several custom metrics.
 
 Incremented once per CLI startup.
 
+##### Onboarding
+
+Tracks onboarding flow from authentication to the user
+
+- `gemini_cli.onboarding.start` (Counter, Int): Incremented when the
+  authentication flow begins.
+
+- `gemini_cli.onboarding.success` (Counter, Int): Incremented when the user
+onboarding flow completes successfully.
+<details>
+<summary>Attributes (Success)</summary>
+
+- `user_tier` (string)
+
 ##### Tools
 
 ##### `gemini_cli.tool.call.count`

docs/cli/themes.md

@@ -36,9 +36,11 @@ using the `/theme` command within Gemini CLI:
     preview or highlight as you select.
 4.  Confirm your selection to apply the theme.
 
-**Note:** If a theme is defined in your `settings.json` file (either by name or
-by a file path), you must remove the `"theme"` setting from the file before you
-can change the theme using the `/theme` command.
+<!-- prettier-ignore -->
+> [!NOTE]
+> If a theme is defined in your `settings.json` file (either by name or
+> by a file path), you must remove the `"theme"` setting from the file before
+> you can change the theme using the `/theme` command.
 
 ### Theme persistence
 
@@ -179,11 +181,13 @@ custom theme defined in `settings.json`.
 }
 ```
 
-**Security note:** For your safety, Gemini CLI will only load theme files that
-are located within your home directory. If you attempt to load a theme from
-outside your home directory, a warning will be displayed and the theme will not
-be loaded. This is to prevent loading potentially malicious theme files from
-untrusted sources.
+<!-- prettier-ignore -->
+> [!WARNING]
+> For your safety, Gemini CLI will only load theme files that
+> are located within your home directory. If you attempt to load a theme from
+> outside your home directory, a warning will be displayed and the theme will
+> not be loaded. This is to prevent loading potentially malicious theme files
+> from untrusted sources.
 
 ### Example custom theme
 

docs/cli/tutorials/file-management.md

@@ -7,9 +7,9 @@ create files, and control what Gemini CLI can see.
 ## Prerequisites
 
 - Gemini CLI installed and authenticated.
-- A project directory to work with (e.g., a git repository).
+- A project directory to work with (for example, a git repository).
 
-## How to give the agent context (Reading files)
+## Providing context by reading files
 
 Gemini CLI will generally try to read relevant files, sometimes prompting you
 for access (depending on your settings). To ensure that Gemini CLI uses a file,
@@ -58,11 +58,13 @@ You know there's a `UserProfile` component, but you don't know where it lives.
 ```
 
 Gemini uses the `glob` or `list_directory` tools to search your project
-structure. It will return the specific path (e.g.,
+structure. It will return the specific path (for example,
 `src/components/UserProfile.tsx`), which you can then use with `@` in your next
 turn.
 
-> **Tip:** You can also ask for lists of files, like "Show me all the TypeScript
+<!-- prettier-ignore -->
+> [!TIP]
+> You can also ask for lists of files, like "Show me all the TypeScript
 > configuration files in the root directory."
 
 ## How to modify code
@@ -111,8 +113,8 @@ or, better yet, run your project's tests.
 `Run the tests for the UserProfile component.`
 ```
 
-Gemini CLI uses the `run_shell_command` tool to execute your test runner (e.g.,
-`npm test` or `jest`). This ensures the changes didn't break existing
+Gemini CLI uses the `run_shell_command` tool to execute your test runner (for
+example, `npm test` or `jest`). This ensures the changes didn't break existing
 functionality.
 
 ## Advanced: Controlling what Gemini sees

docs/cli/tutorials/mcp-setup.md

@@ -62,8 +62,10 @@ You tell Gemini about new servers by editing your `settings.json`.
 }
 ```
 
-> **Note:** The `command` is `docker`, and the rest are arguments passed to it.
-> We map the local environment variable into the container so your secret isn't
+<!-- prettier-ignore -->
+> [!NOTE]
+> The `command` is `docker`, and the rest are arguments passed to it. We
+> map the local environment variable into the container so your secret isn't
 > hardcoded in the config file.
 
 ## How to verify the connection

docs/cli/tutorials/memory-management.md

@@ -11,8 +11,8 @@ persistent facts, and inspect the active context.
 
 ## Why manage context?
 
-Out of the box, Gemini CLI is smart but generic. It doesn't know your preferred
-testing framework, your indentation style, or that you hate using `any` in
+Gemini CLI is powerful but general. It doesn't know your preferred testing
+framework, your indentation style, or your preference against `any` in
 TypeScript. Context management solves this by giving the agent persistent
 memory.
 
@@ -109,11 +109,11 @@ immediately. Force a reload with:
 
 ## Best practices
 
-- **Keep it focused:** Don't dump your entire internal wiki into `GEMINI.md`.
-  Keep instructions actionable and relevant to code generation.
+- **Keep it focused:** Avoid adding excessive content to `GEMINI.md`. Keep
+  instructions actionable and relevant to code generation.
 - **Use negative constraints:** Explicitly telling the agent what _not_ to do
-  (e.g., "Do not use class components") is often more effective than vague
-  positive instructions.
+  (for example, "Do not use class components") is often more effective than
+  vague positive instructions.
 - **Review often:** Periodically check your `GEMINI.md` files to remove outdated
   rules.
 

docs/cli/tutorials/plan-mode-steering.md

@@ -5,9 +5,10 @@ structured environment with model steering's real-time feedback, you can guide
 Gemini CLI through the research and design phases to ensure the final
 implementation plan is exactly what you need.
 
-> **Note:** This is a preview feature under active development. Preview features
-> may only be available in the **Preview** channel or may need to be enabled
-> under `/settings`.
+<!-- prettier-ignore -->
+> [!NOTE]
+> This is an experimental feature currently under active development and
+> may need to be enabled under `/settings`.
 
 ## Prerequisites
 

docs/cli/tutorials/shell-commands.md

@@ -7,7 +7,7 @@ automate complex workflows, and manage background processes safely.
 ## Prerequisites
 
 - Gemini CLI installed and authenticated.
-- Basic familiarity with your system's shell (Bash, Zsh, PowerShell, etc.).
+- Basic familiarity with your system's shell (Bash, Zsh, PowerShell, and so on).
 
 ## How to run commands directly (`!`)
 
@@ -49,7 +49,7 @@ You want to run tests and fix any failures.
 6.  Gemini uses `replace` to fix the bug.
 7.  Gemini runs `npm test` again to verify the fix.
 
-This loop turns Gemini into an autonomous engineer.
+This loop lets Gemini work autonomously.
 
 ## How to manage background processes
 
@@ -75,7 +75,7 @@ confirmation prompts) by streaming the output to you. However, for highly
 interactive tools (like `vim` or `top`), it's often better to run them yourself
 in a separate terminal window or use the `!` prefix.
 
-## Safety first
+## Safety features
 
 Giving an AI access to your shell is powerful but risky. Gemini CLI includes
 several safety layers.

docs/core/remote-agents.md

@@ -10,7 +10,9 @@ agents in the following repositories:
 - [ADK Samples (Python)](https://github.com/google/adk-samples/tree/main/python)
 - [ADK Python Contributing Samples](https://github.com/google/adk-python/tree/main/contributing/samples)
 
-> **Note: Remote subagents are currently an experimental feature.**
+<!-- prettier-ignore -->
+> [!NOTE]
+> Remote subagents are currently an experimental feature.
 
 ## Configuration
 
@@ -82,7 +84,8 @@ Markdown file.
 ---
 ```
 
-> **Note:** Mixed local and remote agents, or multiple local agents, are not
+<!-- prettier-ignore -->
+> [!NOTE] Mixed local and remote agents, or multiple local agents, are not
 > supported in a single file; the list format is currently remote-only.
 
 ## Authentication
@@ -101,7 +104,7 @@ Gemini CLI supports the following authentication types:
 | `apiKey`             | Send a static API key as an HTTP header.                                                       |
 | `http`               | HTTP authentication (Bearer token, Basic credentials, or any IANA-registered scheme).          |
 | `google-credentials` | Google Application Default Credentials (ADC). Automatically selects access or identity tokens. |
-| `oauth2`             | OAuth 2.0 Authorization Code flow with PKCE. Opens a browser for interactive sign-in.          |
+| `oauth`              | OAuth 2.0 Authorization Code flow with PKCE. Opens a browser for interactive sign-in.          |
 
 ### Dynamic values
 
@@ -260,7 +263,7 @@ hosts:
 
 Requests to any other host will be rejected with an error. If your agent is
 hosted on a different domain, use one of the other auth types (`apiKey`, `http`,
-or `oauth2`).
+or `oauth`).
 
 #### Examples
 
@@ -294,7 +297,7 @@ auth:
 ---
 ```
 
-### OAuth 2.0 (`oauth2`)
+### OAuth 2.0 (`oauth`)
 
 Performs an interactive OAuth 2.0 Authorization Code flow with PKCE. On first
 use, Gemini CLI opens your browser for sign-in and persists the resulting tokens
@@ -302,7 +305,7 @@ for subsequent requests.
 
 | Field               | Type     | Required | Description                                                                                                                                        |
 | :------------------ | :------- | :------- | :------------------------------------------------------------------------------------------------------------------------------------------------- |
-| `type`              | string   | Yes      | Must be `oauth2`.                                                                                                                                  |
+| `type`              | string   | Yes      | Must be `oauth`.                                                                                                                                   |
 | `client_id`         | string   | Yes\*    | OAuth client ID. Required for interactive auth.                                                                                                    |
 | `client_secret`     | string   | No\*     | OAuth client secret. Required by most authorization servers (confidential clients). Can be omitted for public clients that don't require a secret. |
 | `scopes`            | string[] | No       | Requested scopes. Can also be discovered from the agent card.                                                                                      |
@@ -315,7 +318,7 @@ kind: remote
 name: oauth-agent
 agent_card_url: https://example.com/.well-known/agent.json
 auth:
-  type: oauth2
+  type: oauth
   client_id: my-client-id.apps.example.com
 ---
 ```
@@ -362,5 +365,7 @@ Users can manage subagents using the following commands within the Gemini CLI:
 - `/agents enable <agent_name>`: Enables a specific subagent.
 - `/agents disable <agent_name>`: Disables a specific subagent.
 
-> **Tip:** You can use the `@cli_help` agent within Gemini CLI for assistance
+<!-- prettier-ignore -->
+> [!TIP]
+> You can use the `@cli_help` agent within Gemini CLI for assistance
 > with configuring subagents.

docs/core/subagents.md

@@ -5,16 +5,18 @@ session. They are designed to handle specific, complex tasks—like deep codebas
 analysis, documentation lookup, or domain-specific reasoning—without cluttering
 the main agent's context or toolset.
 
-> **Note: Subagents are currently an experimental feature.**
->
-> To use custom subagents, you must ensure they are enabled in your
-> `settings.json` (enabled by default):
->
-> ```json
-> {
->   "experimental": { "enableAgents": true }
-> }
-> ```
+<!-- prettier-ignore -->
+> [!NOTE]
+> Subagents are currently an experimental feature.
+> 
+To use custom subagents, you must ensure they are enabled in your
+`settings.json` (enabled by default):
+
+```json
+{
+  "experimental": { "enableAgents": true }
+}
+```
 
 ## What are subagents?
 
@@ -114,7 +116,9 @@ Gemini CLI comes with the following built-in subagents:
   the pricing table from this page," "Click the login button and enter my
   credentials."
 
-> **Note:** This is a preview feature currently under active development.
+<!-- prettier-ignore -->
+> [!NOTE]
+> This is a preview feature currently under active development.
 
 #### Prerequisites
 
@@ -217,7 +221,9 @@ captures a screenshot and sends it to the vision model for analysis. The model
 returns coordinates and element descriptions that the browser agent uses with
 the `click_at` tool for precise, coordinate-based interactions.
 
-> **Note:** The visual agent requires API key or Vertex AI authentication. It is
+<!-- prettier-ignore -->
+> [!NOTE]
+> The visual agent requires API key or Vertex AI authentication. It is
 > not available when using "Sign in with Google".
 
 ## Creating custom subagents
@@ -405,7 +411,9 @@ that your subagent was called with a specific prompt and the given description.
 Gemini CLI can also delegate tasks to remote subagents using the Agent-to-Agent
 (A2A) protocol.
 
-> **Note: Remote subagents are currently an experimental feature.**
+<!-- prettier-ignore -->
+> [!NOTE]
+> Remote subagents are currently an experimental feature.
 
 See the [Remote Subagents documentation](remote-agents) for detailed
 configuration, authentication, and usage instructions.

docs/extensions/reference.md

@@ -23,7 +23,7 @@ Gemini CLI creates a copy of the extension during installation. You must run
 GitHub, you must have `git` installed on your machine.
 
 ```bash
-gemini extensions install <source> [--ref <ref>] [--auto-update] [--pre-release] [--consent]
+gemini extensions install <source> [--ref <ref>] [--auto-update] [--pre-release] [--consent] [--skip-settings]
 ```
 
 - `<source>`: The GitHub URL or local path of the extension.
@@ -31,6 +31,7 @@ gemini extensions install <source> [--ref <ref>] [--auto-update] [--pre-release]
 - `--auto-update`: Enable automatic updates for this extension.
 - `--pre-release`: Enable installation of pre-release versions.
 - `--consent`: Acknowledge security risks and skip the confirmation prompt.
+- `--skip-settings`: Skip the configuration on install process.
 
 ### Uninstall an extension
 
@@ -234,7 +235,9 @@ skill definitions in a `skills/` directory. For example,
 
 ### Sub-agents
 
-> **Note:** Sub-agents are a preview feature currently under active development.
+<!-- prettier-ignore -->
+> [!NOTE]
+> Sub-agents are a preview feature currently under active development.
 
 Provide [sub-agents](../core/subagents.md) that users can delegate tasks to. Add
 agent definition files (`.md`) to an `agents/` directory in your extension root.
@@ -253,7 +256,9 @@ Rules contributed by extensions run in their own tier (tier 2), alongside
 workspace-defined policies. This tier has higher priority than the default rules
 but lower priority than user or admin policies.
 
-> **Warning:** For security, Gemini CLI ignores any `allow` decisions or `yolo`
+<!-- prettier-ignore -->
+> [!WARNING]
+> For security, Gemini CLI ignores any `allow` decisions or `yolo`
 > mode configurations in extension policies. This ensures that an extension
 > cannot automatically approve tool calls or bypass security measures without
 > your confirmation.

docs/get-started/authentication.md

@@ -4,7 +4,9 @@ To use Gemini CLI, you'll need to authenticate with Google. This guide helps you
 quickly find the best way to sign in based on your account type and how you're
 using the CLI.
 
-> **Note:** Looking for a high-level comparison of all available subscriptions?
+<!-- prettier-ignore -->
+> [!TIP]
+> Looking for a high-level comparison of all available subscriptions?
 > To compare features and find the right quota for your needs, see our
 > [Plans page](https://geminicli.com/plans/).
 
@@ -40,11 +42,11 @@ Select the authentication method that matches your situation in the table below:
 
 If you run Gemini CLI on your local machine, the simplest authentication method
 is logging in with your Google account. This method requires a web browser on a
-machine that can communicate with the terminal running Gemini CLI (e.g., your
-local machine).
+machine that can communicate with the terminal running Gemini CLI (for example,
+your local machine).
 
-> **Important:** If you are a **Google AI Pro** or **Google AI Ultra**
-> subscriber, use the Google account associated with your subscription.
+If you are a **Google AI Pro** or **Google AI Ultra** subscriber, use the Google
+account associated with your subscription.
 
 To authenticate and use Gemini CLI:
 
@@ -107,7 +109,9 @@ To authenticate and use Gemini CLI with a Gemini API key:
 
 4. Select **Use Gemini API key**.
 
-> **Warning:** Treat API keys, especially for services like Gemini, as sensitive
+<!-- prettier-ignore -->
+> [!WARNING]
+> Treat API keys, especially for services like Gemini, as sensitive
 > credentials. Protect them to prevent unauthorized access and potential misuse
 > of the service under your account.
 
@@ -130,7 +134,7 @@ For example:
 **macOS/Linux**
 
 ```bash
-# Replace with your project ID and desired location (e.g., us-central1)
+# Replace with your project ID and desired location (for example, us-central1)
 export GOOGLE_CLOUD_PROJECT="YOUR_PROJECT_ID"
 export GOOGLE_CLOUD_LOCATION="YOUR_PROJECT_LOCATION"
 ```
@@ -138,7 +142,7 @@ export GOOGLE_CLOUD_LOCATION="YOUR_PROJECT_LOCATION"
 **Windows (PowerShell)**
 
 ```powershell
-# Replace with your project ID and desired location (e.g., us-central1)
+# Replace with your project ID and desired location (for example, us-central1)
 $env:GOOGLE_CLOUD_PROJECT="YOUR_PROJECT_ID"
 $env:GOOGLE_CLOUD_LOCATION="YOUR_PROJECT_LOCATION"
 ```
@@ -150,20 +154,20 @@ To make any Vertex AI environment variable settings persistent, see
 
 Consider this authentication method if you have Google Cloud CLI installed.
 
-> **Note:** If you have previously set `GOOGLE_API_KEY` or `GEMINI_API_KEY`, you
-> must unset them to use ADC:
->
-> **macOS/Linux**
->
-> ```bash
-> unset GOOGLE_API_KEY GEMINI_API_KEY
-> ```
->
-> **Windows (PowerShell)**
->
-> ```powershell
-> Remove-Item Env:\GOOGLE_API_KEY, Env:\GEMINI_API_KEY -ErrorAction Ignore
-> ```
+If you have previously set `GOOGLE_API_KEY` or `GEMINI_API_KEY`, you must unset
+them to use ADC.
+
+**macOS/Linux**
+
+```bash
+unset GOOGLE_API_KEY GEMINI_API_KEY
+```
+
+**Windows (PowerShell)**
+
+```powershell
+Remove-Item Env:\GOOGLE_API_KEY, Env:\GEMINI_API_KEY -ErrorAction Ignore
+```
 
 1. Verify you have a Google Cloud project and Vertex AI API is enabled.
 
@@ -188,20 +192,20 @@ Consider this authentication method if you have Google Cloud CLI installed.
 Consider this method of authentication in non-interactive environments, CI/CD
 pipelines, or if your organization restricts user-based ADC or API key creation.
 
-> **Note:** If you have previously set `GOOGLE_API_KEY` or `GEMINI_API_KEY`, you
-> must unset them:
->
-> **macOS/Linux**
->
-> ```bash
-> unset GOOGLE_API_KEY GEMINI_API_KEY
-> ```
->
-> **Windows (PowerShell)**
->
-> ```powershell
-> Remove-Item Env:\GOOGLE_API_KEY, Env:\GEMINI_API_KEY -ErrorAction Ignore
-> ```
+If you have previously set `GOOGLE_API_KEY` or `GEMINI_API_KEY`, you must unset
+them:
+
+**macOS/Linux**
+
+```bash
+unset GOOGLE_API_KEY GEMINI_API_KEY
+```
+
+**Windows (PowerShell)**
+
+```powershell
+Remove-Item Env:\GOOGLE_API_KEY, Env:\GEMINI_API_KEY -ErrorAction Ignore
+```
 
 1.  [Create a service account and key](https://cloud.google.com/iam/docs/keys-create-delete)
     and download the provided JSON file. Assign the "Vertex AI User" role to the
@@ -233,8 +237,11 @@ pipelines, or if your organization restricts user-based ADC or API key creation.
     ```
 
 5.  Select **Vertex AI**.
-    > **Warning:** Protect your service account key file as it gives access to
-    > your resources.
+
+<!-- prettier-ignore -->
+> [!WARNING]
+> Protect your service account key file as it gives access to
+> your resources.
 
 #### C. Vertex AI - Google Cloud API key
 
@@ -257,10 +264,9 @@ pipelines, or if your organization restricts user-based ADC or API key creation.
     $env:GOOGLE_API_KEY="YOUR_GOOGLE_API_KEY"
     ```
 
-    > **Note:** If you see errors like
-    > `"API keys are not supported by this API..."`, your organization might
-    > restrict API key usage for this service. Try the other Vertex AI
-    > authentication methods instead.
+    If you see errors like `"API keys are not supported by this API..."`, your
+    organization might restrict API key usage for this service. Try the other
+    Vertex AI authentication methods instead.
 
 3.  [Configure your Google Cloud Project](#set-gcp).
 
@@ -274,7 +280,9 @@ pipelines, or if your organization restricts user-based ADC or API key creation.
 
 ## Set your Google Cloud project <a id="set-gcp"></a>
 
-> **Important:** Most individual Google accounts (free and paid) don't require a
+<!-- prettier-ignore -->
+> [!IMPORTANT]
+> Most individual Google accounts (free and paid) don't require a
 > Google Cloud project for authentication.
 
 When you sign in using your Google account, you may need to configure a Google
@@ -325,29 +333,31 @@ persist them with the following methods:
 1.  **Add your environment variables to your shell configuration file:** Append
     the environment variable commands to your shell's startup file.
 
-    **macOS/Linux** (e.g., `~/.bashrc`, `~/.zshrc`, or `~/.profile`):
+    **macOS/Linux** (for example, `~/.bashrc`, `~/.zshrc`, or `~/.profile`):
 
     ```bash
     echo 'export GOOGLE_CLOUD_PROJECT="YOUR_PROJECT_ID"' >> ~/.bashrc
     source ~/.bashrc
     ```
 
-    **Windows (PowerShell)** (e.g., `$PROFILE`):
+    **Windows (PowerShell)** (for example, `$PROFILE`):
 
     ```powershell
     Add-Content -Path $PROFILE -Value '$env:GOOGLE_CLOUD_PROJECT="YOUR_PROJECT_ID"'
     . $PROFILE
     ```
 
-    > **Warning:** Be aware that when you export API keys or service account
-    > paths in your shell configuration file, any process launched from that
-    > shell can read them.
+<!-- prettier-ignore -->
+> [!WARNING]
+> Be aware that when you export API keys or service account
+> paths in your shell configuration file, any process launched from that
+> shell can read them.
 
 2.  **Use a `.env` file:** Create a `.gemini/.env` file in your project
     directory or home directory. Gemini CLI automatically loads variables from
     the first `.env` file it finds, searching up from the current directory,
-    then in your home directory's `.gemini/.env` (e.g., `~/.gemini/.env` or
-    `%USERPROFILE%\.gemini\.env`).
+    then in your home directory's `.gemini/.env` (for example, `~/.gemini/.env`
+    or `%USERPROFILE%\.gemini\.env`).
 
     Example for user-wide settings:
 

docs/get-started/examples.md

@@ -4,7 +4,9 @@ Gemini CLI helps you automate common engineering tasks by combining AI reasoning
 with local system tools. This document provides examples of how to use the CLI
 for file management, code analysis, and data transformation.
 
-> **Note:** These examples demonstrate potential capabilities. Your actual
+<!-- prettier-ignore -->
+> [!NOTE]
+> These examples demonstrate potential capabilities. Your actual
 > results can vary based on the model used and your project environment.
 
 ## Rename your photographs based on content

docs/get-started/gemini-3.md

@@ -2,7 +2,9 @@
 
 Gemini 3 Pro and Gemini 3 Flash are available on Gemini CLI for all users!
 
-> **Note:** Gemini 3.1 Pro Preview is rolling out. To determine whether you have
+<!-- prettier-ignore -->
+> [!NOTE]
+> Gemini 3.1 Pro Preview is rolling out. To determine whether you have
 > access to Gemini 3.1, use the `/model` command and select **Manual**. If you
 > have access, you will see `gemini-3.1-pro-preview`.
 >
@@ -25,7 +27,7 @@ Get started by upgrading Gemini CLI to the latest version:
 npm install -g @google/gemini-cli@latest
 ```
 
-After you’ve confirmed your version is 0.21.1 or later:
+If your version is 0.21.1 or later:
 
 1. Run `/model`.
 2. Select **Auto (Gemini 3)**.
@@ -39,7 +41,9 @@ When you encounter that limit, you’ll be given the option to switch to Gemini
 2.5 Pro, upgrade for higher limits, or stop. You’ll also be told when your usage
 limit resets and Gemini 3 Pro can be used again.
 
-> **Note:** Looking to upgrade for higher limits? To compare subscription
+<!-- prettier-ignore -->
+> [!TIP]
+> Looking to upgrade for higher limits? To compare subscription
 > options and find the right quota for your needs, see our
 > [Plans page](https://geminicli.com/plans/).
 
@@ -52,7 +56,9 @@ There may be times when the Gemini 3 Pro model is overloaded. When that happens,
 Gemini CLI will ask you to decide whether you want to keep trying Gemini 3 Pro
 or fallback to Gemini 2.5 Pro.
 
-> **Note:** The **Keep trying** option uses exponential backoff, in which Gemini
+<!-- prettier-ignore -->
+> [!NOTE]
+> The **Keep trying** option uses exponential backoff, in which Gemini
 > CLI waits longer between each retry, when the system is busy. If the retry
 > doesn't happen immediately, please wait a few minutes for the request to
 > process.
@@ -109,7 +115,7 @@ then:
 
 Restart Gemini CLI and you should have access to Gemini 3.
 
-## Need help?
+## Next steps
 
 If you need help, we recommend searching for an existing
 [GitHub issue](https://github.com/google-gemini/gemini-cli/issues). If you

docs/hooks/index.md

@@ -143,7 +143,9 @@ Hooks are executed with a sanitized environment.
 
 ## Security and risks
 
-> **Warning: Hooks execute arbitrary code with your user privileges.** By
+<!-- prettier-ignore -->
+> [!WARNING]
+> Hooks execute arbitrary code with your user privileges. By
 > configuring hooks, you are allowing scripts to run shell commands on your
 > machine.
 

docs/hooks/writing-hooks.md

@@ -470,5 +470,5 @@ console.error('Consolidating memories for session end...');
 
 While project-level hooks are great for specific repositories, you can share
 your hooks across multiple projects by packaging them as a
-[Gemini CLI extension](https://www.google.com/search?q=../extensions/index.md).
-This provides version control, easy distribution, and centralized management.
+[Gemini CLI extension](../extensions/index.md). This provides version control,
+easy distribution, and centralized management.

docs/ide-integration/ide-companion-spec.md

@@ -132,9 +132,11 @@ to the CLI whenever the user's context changes.
   }
   ```
 
-  **Note:** The `openFiles` list should only include files that exist on disk.
-  Virtual files (e.g., unsaved files without a path, editor settings pages)
-  **MUST** be excluded.
+<!-- prettier-ignore -->
+> [!NOTE]
+> The `openFiles` list should only include files that exist on disk.
+> Virtual files (e.g., unsaved files without a path, editor settings pages)
+> **MUST** be excluded.
 
 ### How the CLI uses this context
 

docs/ide-integration/index.md

@@ -66,9 +66,11 @@ You can also install the extension directly from a marketplace.
   Follow your editor's instructions for installing extensions from this
   registry.
 
-> NOTE: The "Gemini CLI Companion" extension may appear towards the bottom of
-> search results. If you don't see it immediately, try scrolling down or sorting
-> by "Newly Published".
+<!-- prettier-ignore -->
+> [!NOTE]
+> The "Gemini CLI Companion" extension may appear towards the bottom of
+> search results. If you don't see it immediately, try scrolling down or
+> sorting by "Newly Published".
 >
 > After manually installing the extension, you must run `/ide enable` in the CLI
 > to activate the integration.
@@ -103,7 +105,9 @@ IDE, run:
 If connected, this command will show the IDE it's connected to and a list of
 recently opened files it is aware of.
 
-> [!NOTE] The file list is limited to 10 recently accessed files within your
+<!-- prettier-ignore -->
+> [!NOTE]
+> The file list is limited to 10 recently accessed files within your
 > workspace and only includes local files on disk.)
 
 ### Working with diffs

docs/issue-and-pr-automation.md

@@ -14,7 +14,9 @@ feature), while the PR is the "how" (the implementation). This separation helps
 us track work, prioritize features, and maintain clear historical context. Our
 automation is built around this principle.
 
-> **Note:** Issues tagged as "🔒Maintainers only" are reserved for project
+<!-- prettier-ignore -->
+> [!NOTE]
+> Issues tagged as "🔒Maintainers only" are reserved for project
 > maintainers. We will not accept pull requests related to these issues.
 
 ---

docs/local-development.md

@@ -79,7 +79,9 @@ You can view traces in the Jaeger UI for local development.
 You can use an OpenTelemetry collector to forward telemetry data to Google Cloud
 Trace for custom processing or routing.
 
-> **Warning:** Ensure you complete the
+<!-- prettier-ignore -->
+> [!WARNING]
+> Ensure you complete the
 > [Google Cloud telemetry prerequisites](./cli/telemetry.md#prerequisites)
 > (Project ID, authentication, IAM roles, and APIs) before using this method.
 

docs/reference/commands.md

@@ -60,8 +60,8 @@ Slash commands provide meta-level control over the CLI itself.
     - `list` (selecting this opens the auto-saved session browser)
   - `-- checkpoints --`
     - `list`, `save`, `resume`, `delete`, `share` (manual tagged checkpoints)
-  - **Note:** Unique prefixes (for example `/cha` or `/resum`) resolve to the
-    same grouped menu.
+  - Unique prefixes (for example `/cha` or `/resu`) resolve to the same grouped
+    menu.
 - **Sub-commands:**
   - **`debug`**
     - **Description:** Export the most recent API request as a JSON payload.
@@ -250,8 +250,8 @@ Slash commands provide meta-level control over the CLI itself.
   - **`list`** or **`ls`**:
     - **Description:** List configured MCP servers and tools. This is the
       default action if no subcommand is specified.
-  - **`refresh`**:
-    - **Description:** Restarts all MCP servers and re-discovers their available
+  - **`reload`**:
+    - **Description:** Reloads all MCP servers and re-discovers their available
       tools.
   - **`schema`**:
     - **Description:** List configured MCP servers and tools with descriptions

docs/reference/configuration.md

@@ -25,7 +25,9 @@ overridden by higher numbers):
 Gemini CLI uses JSON settings files for persistent configuration. There are four
 locations for these files:
 
-> **Tip:** JSON-aware editors can use autocomplete and validation by pointing to
+<!-- prettier-ignore -->
+> [!TIP]
+> JSON-aware editors can use autocomplete and validation by pointing to
 > the generated schema at `schemas/settings.schema.json` in this repository.
 > When working outside the repo, reference the hosted schema at
 > `https://raw.githubusercontent.com/google-gemini/gemini-cli/main/schemas/settings.schema.json`.
@@ -66,9 +68,9 @@ an environment variable `MY_API_TOKEN`, you could use it in `settings.json` like
 this: `"apiKey": "$MY_API_TOKEN"`. Additionally, each extension can have its own
 `.env` file in its directory, which will be loaded automatically.
 
-> **Note for Enterprise Users:** For guidance on deploying and managing Gemini
-> CLI in a corporate environment, please see the
-> [Enterprise Configuration](../cli/enterprise.md) documentation.
+**Note for Enterprise Users:** For guidance on deploying and managing Gemini CLI
+in a corporate environment, please see the
+[Enterprise Configuration](../cli/enterprise.md) documentation.
 
 ### The `.gemini` directory in your project
 
@@ -293,6 +295,11 @@ their corresponding top-level category object in your `settings.json` file.
   - **Description:** Hide the footer from the UI
   - **Default:** `false`
 
+- **`ui.collapseDrawerDuringApproval`** (boolean):
+  - **Description:** Whether to collapse the UI drawer when a tool is awaiting
+    confirmation.
+  - **Default:** `true`
+
 - **`ui.showMemoryUsage`** (boolean):
   - **Description:** Display memory usage information in the UI
   - **Default:** `false`
@@ -684,6 +691,16 @@ their corresponding top-level category object in your `settings.json` file.
 
     ```json
     {
+      "gemini-3.1-flash-lite-preview": {
+        "tier": "flash-lite",
+        "family": "gemini-3",
+        "isPreview": true,
+        "isVisible": true,
+        "features": {
+          "thinking": false,
+          "multimodalToolUse": true
+        }
+      },
       "gemini-3.1-pro-preview": {
         "tier": "pro",
         "family": "gemini-3",
@@ -795,7 +812,7 @@ their corresponding top-level category object in your `settings.json` file.
         "tier": "auto",
         "isPreview": true,
         "isVisible": true,
-        "dialogDescription": "Let Gemini CLI decide the best model for the task: gemini-3.1-pro, gemini-3-flash",
+        "dialogDescription": "Let Gemini CLI decide the best model for the task: gemini-3-pro, gemini-3-flash",
         "features": {
           "thinking": true,
           "multimodalToolUse": false
@@ -824,6 +841,45 @@ their corresponding top-level category object in your `settings.json` file.
 
     ```json
     {
+      "gemini-3.1-pro-preview": {
+        "default": "gemini-3.1-pro-preview",
+        "contexts": [
+          {
+            "condition": {
+              "hasAccessToPreview": false
+            },
+            "target": "gemini-2.5-pro"
+          },
+          {
+            "condition": {
+              "useCustomTools": true
+            },
+            "target": "gemini-3.1-pro-preview-customtools"
+          }
+        ]
+      },
+      "gemini-3.1-pro-preview-customtools": {
+        "default": "gemini-3.1-pro-preview-customtools",
+        "contexts": [
+          {
+            "condition": {
+              "hasAccessToPreview": false
+            },
+            "target": "gemini-2.5-pro"
+          }
+        ]
+      },
+      "gemini-3-flash-preview": {
+        "default": "gemini-3-flash-preview",
+        "contexts": [
+          {
+            "condition": {
+              "hasAccessToPreview": false
+            },
+            "target": "gemini-2.5-flash"
+          }
+        ]
+      },
       "gemini-3-pro-preview": {
         "default": "gemini-3-pro-preview",
         "contexts": [
@@ -995,6 +1051,132 @@ their corresponding top-level category object in your `settings.json` file.
 
   - **Requires restart:** Yes
 
+- **`modelConfigs.modelChains`** (object):
+  - **Description:** Availability policy chains defining fallback behavior for
+    models.
+  - **Default:**
+
+    ```json
+    {
+      "preview": [
+        {
+          "model": "gemini-3-pro-preview",
+          "actions": {
+            "terminal": "prompt",
+            "transient": "prompt",
+            "not_found": "prompt",
+            "unknown": "prompt"
+          },
+          "stateTransitions": {
+            "terminal": "terminal",
+            "transient": "terminal",
+            "not_found": "terminal",
+            "unknown": "terminal"
+          }
+        },
+        {
+          "model": "gemini-3-flash-preview",
+          "isLastResort": true,
+          "actions": {
+            "terminal": "prompt",
+            "transient": "prompt",
+            "not_found": "prompt",
+            "unknown": "prompt"
+          },
+          "stateTransitions": {
+            "terminal": "terminal",
+            "transient": "terminal",
+            "not_found": "terminal",
+            "unknown": "terminal"
+          }
+        }
+      ],
+      "default": [
+        {
+          "model": "gemini-2.5-pro",
+          "actions": {
+            "terminal": "prompt",
+            "transient": "prompt",
+            "not_found": "prompt",
+            "unknown": "prompt"
+          },
+          "stateTransitions": {
+            "terminal": "terminal",
+            "transient": "terminal",
+            "not_found": "terminal",
+            "unknown": "terminal"
+          }
+        },
+        {
+          "model": "gemini-2.5-flash",
+          "isLastResort": true,
+          "actions": {
+            "terminal": "prompt",
+            "transient": "prompt",
+            "not_found": "prompt",
+            "unknown": "prompt"
+          },
+          "stateTransitions": {
+            "terminal": "terminal",
+            "transient": "terminal",
+            "not_found": "terminal",
+            "unknown": "terminal"
+          }
+        }
+      ],
+      "lite": [
+        {
+          "model": "gemini-2.5-flash-lite",
+          "actions": {
+            "terminal": "silent",
+            "transient": "silent",
+            "not_found": "silent",
+            "unknown": "silent"
+          },
+          "stateTransitions": {
+            "terminal": "terminal",
+            "transient": "terminal",
+            "not_found": "terminal",
+            "unknown": "terminal"
+          }
+        },
+        {
+          "model": "gemini-2.5-flash",
+          "actions": {
+            "terminal": "silent",
+            "transient": "silent",
+            "not_found": "silent",
+            "unknown": "silent"
+          },
+          "stateTransitions": {
+            "terminal": "terminal",
+            "transient": "terminal",
+            "not_found": "terminal",
+            "unknown": "terminal"
+          }
+        },
+        {
+          "model": "gemini-2.5-pro",
+          "isLastResort": true,
+          "actions": {
+            "terminal": "silent",
+            "transient": "silent",
+            "not_found": "silent",
+            "unknown": "silent"
+          },
+          "stateTransitions": {
+            "terminal": "terminal",
+            "transient": "terminal",
+            "not_found": "terminal",
+            "unknown": "terminal"
+          }
+        }
+      ]
+    }
+    ```
+
+  - **Requires restart:** Yes
+
 #### `agents`
 
 - **`agents.overrides`** (object):
@@ -1039,6 +1221,22 @@ their corresponding top-level category object in your `settings.json` file.
   - **Description:** Disable user input on browser window during automation.
   - **Default:** `true`
 
+- **`agents.browser.maxActionsPerTask`** (number):
+  - **Description:** The maximum number of tool calls allowed per browser task.
+    Enforcement is hard: the agent will be terminated when the limit is reached.
+  - **Default:** `100`
+
+- **`agents.browser.confirmSensitiveActions`** (boolean):
+  - **Description:** Require manual confirmation for sensitive browser actions
+    (e.g., fill_form, evaluate_script).
+  - **Default:** `false`
+  - **Requires restart:** Yes
+
+- **`agents.browser.blockFileUploads`** (boolean):
+  - **Description:** Hard-block file upload requests from the browser agent.
+  - **Default:** `false`
+  - **Requires restart:** Yes
+
 #### `context`
 
 - **`context.fileName`** (string | string[]):
@@ -1105,10 +1303,21 @@ their corresponding top-level category object in your `settings.json` file.
   - **Description:** Legacy full-process sandbox execution environment. Set to a
     boolean to enable or disable the sandbox, provide a string path to a sandbox
     profile, or specify an explicit sandbox command (e.g., "docker", "podman",
-    "lxc").
+    "lxc", "windows-native").
   - **Default:** `undefined`
   - **Requires restart:** Yes
 
+- **`tools.sandboxAllowedPaths`** (array):
+  - **Description:** List of additional paths that the sandbox is allowed to
+    access.
+  - **Default:** `[]`
+  - **Requires restart:** Yes
+
+- **`tools.sandboxNetworkAccess`** (boolean):
+  - **Description:** Whether the sandbox is allowed to access the network.
+  - **Default:** `false`
+  - **Requires restart:** Yes
+
 - **`tools.shell.enableInteractiveShell`** (boolean):
   - **Description:** Use node-pty for an interactive shell experience. Fallback
     to child_process still applies.
@@ -1345,6 +1554,11 @@ their corresponding top-level category object in your `settings.json` file.
   - **Default:** `true`
   - **Requires restart:** Yes
 
+- **`experimental.worktrees`** (boolean):
+  - **Description:** Enable automated Git worktree management for parallel work.
+  - **Default:** `false`
+  - **Requires restart:** Yes
+
 - **`experimental.extensionManagement`** (boolean):
   - **Description:** Enable extension management features.
   - **Default:** `true`
@@ -1431,6 +1645,13 @@ their corresponding top-level category object in your `settings.json` file.
   - **Default:** `"gemma3-1b-gpu-custom"`
   - **Requires restart:** Yes
 
+- **`experimental.memoryManager`** (boolean):
+  - **Description:** Replace the built-in save_memory tool with a memory manager
+    subagent that supports adding, removing, de-duplicating, and organizing
+    memories.
+  - **Default:** `false`
+  - **Requires restart:** Yes
+
 - **`experimental.topicUpdateNarration`** (boolean):
   - **Description:** Enable the experimental Topic & Update communication model
     for reduced chattiness and structured progress reporting.
@@ -1539,7 +1760,11 @@ their corresponding top-level category object in your `settings.json` file.
   - **Default:** `true`
 
 - **`admin.mcp.config`** (object):
-  - **Description:** Admin-configured MCP servers.
+  - **Description:** Admin-configured MCP servers (allowlist).
+  - **Default:** `{}`
+
+- **`admin.mcp.requiredConfig`** (object):
+  - **Description:** Admin-required MCP servers that are always injected.
   - **Default:** `{}`
 
 - **`admin.skills.enabled`** (boolean):
@@ -1559,7 +1784,9 @@ for compatibility. At least one of `command`, `url`, or `httpUrl` must be
 provided. If multiple are specified, the order of precedence is `httpUrl`, then
 `url`, then `command`.
 
-> **Warning:** Avoid using underscores (`_`) in your server aliases (e.g., use
+<!-- prettier-ignore -->
+> [!WARNING]
+> Avoid using underscores (`_`) in your server aliases (e.g., use
 > `my-server` instead of `my_server`). The underlying policy engine parses Fully
 > Qualified Names (`mcp_server_tool`) using the first underscore after the
 > `mcp_` prefix. An underscore in your server alias will cause the parser to

docs/reference/policy-engine.md

@@ -113,7 +113,9 @@ There are three possible decisions a rule can enforce:
 - `ask_user`: The user is prompted to approve or deny the tool call. (In
   non-interactive mode, this is treated as `deny`.)
 
-> **Note:** The `deny` decision is the recommended way to exclude tools. The
+<!-- prettier-ignore -->
+> [!NOTE]
+> The `deny` decision is the recommended way to exclude tools. The
 > legacy `tools.exclude` setting in `settings.json` is deprecated in favor of
 > policy rules with a `deny` decision.
 
@@ -239,15 +241,17 @@ directory are **ignored**.
 - **Linux / macOS:** Must be owned by `root` (UID 0) and NOT writable by group
   or others (e.g., `chmod 755`).
 - **Windows:** Must be in `C:\ProgramData`. Standard users (`Users`, `Everyone`)
-  must NOT have `Write`, `Modify`, or `Full Control` permissions. _Tip: If you
-  see a security warning, use the folder properties to remove write permissions
-  for non-admin groups. You may need to "Disable inheritance" in Advanced
-  Security Settings._
+  must NOT have `Write`, `Modify`, or `Full Control` permissions. If you see a
+  security warning, use the folder properties to remove write permissions for
+  non-admin groups. You may need to "Disable inheritance" in Advanced Security
+  Settings.
 
-**Note:** Supplemental admin policies (provided via `--admin-policy` or
-`adminPolicyPaths` settings) are **NOT** subject to these strict ownership
-checks, as they are explicitly provided by the user or administrator in their
-current execution context.
+<!-- prettier-ignore -->
+> [!NOTE]
+> Supplemental admin policies (provided via `--admin-policy` or
+> `adminPolicyPaths` settings) are **NOT** subject to these strict ownership
+> checks, as they are explicitly provided by the user or administrator in their
+> current execution context.
 
 ### TOML rule schema
 
@@ -258,8 +262,8 @@ Here is a breakdown of the fields available in a TOML policy rule:
 # A unique name for the tool, or an array of names.
 toolName = "run_shell_command"
 
-# (Optional) The name of a subagent. If provided, the rule only applies to tool calls
-# made by this specific subagent.
+# (Optional) The name of a subagent. If provided, the rule only applies to tool
+# calls made by this specific subagent.
 subagent = "generalist"
 
 # (Optional) The name of an MCP server. Can be combined with toolName
@@ -274,14 +278,17 @@ toolAnnotations = { readOnlyHint = true }
 argsPattern = '"command":"(git|npm)'
 
 # (Optional) A string or array of strings that a shell command must start with.
-# This is syntactic sugar for `toolName = "run_shell_command"` and an `argsPattern`.
+# This is syntactic sugar for `toolName = "run_shell_command"` and an
+# `argsPattern`.
 commandPrefix = "git"
 
 # (Optional) A regex to match against the entire shell command.
 # This is also syntactic sugar for `toolName = "run_shell_command"`.
-# Note: This pattern is tested against the JSON representation of the arguments (e.g., `{"command":"<your_command>"}`).
-# Because it prepends `"command":"`, it effectively matches from the start of the command.
-# Anchors like `^` or `$` apply to the full JSON string, so `^` should usually be avoided here.
+# Note: This pattern is tested against the JSON representation of the arguments
+# (e.g., `{"command":"<your_command>"}`). Because it prepends `"command":"`,
+# it effectively matches from the start of the command.
+# Anchors like `^` or `$` apply to the full JSON string,
+# so `^` should usually be avoided here.
 # You cannot use commandPrefix and commandRegex in the same rule.
 commandRegex = "git (commit|push)"
 
@@ -291,16 +298,26 @@ decision = "ask_user"
 # The priority of the rule, from 0 to 999.
 priority = 10
 
-# (Optional) A custom message to display when a tool call is denied by this rule.
-# This message is returned to the model and user, useful for explaining *why* it was denied.
-deny_message = "Deletion is permanent"
+# (Optional) A custom message to display when a tool call is denied by this
+# rule. This message is returned to the model and user,
+# useful for explaining *why* it was denied.
+denyMessage = "Deletion is permanent"
 
 # (Optional) An array of approval modes where this rule is active.
 modes = ["autoEdit"]
 
-# (Optional) A boolean to restrict the rule to interactive (true) or non-interactive (false) environments.
+# (Optional) A boolean to restrict the rule to interactive (true) or
+# non-interactive (false) environments.
 # If omitted, the rule applies to both.
 interactive = true
+
+# (Optional) If true, lets shell commands use redirection operators
+# (>, >>, <, <<, <<<). By default, the policy engine asks for confirmation
+# when redirection is detected, even if a rule matches the command.
+# This permission is granular; it only applies to the specific rule it's
+# defined in. In chained commands (e.g., cmd1 > file && cmd2), each
+# individual command rule must permit redirection if it's used.
+allowRedirection = true
 ```
 
 ### Using arrays (lists)
@@ -348,7 +365,9 @@ using the `mcpName` field. **This is the recommended approach** for defining MCP
 policies, as it is much more robust than manually writing Fully Qualified Names
 (FQNs) or string wildcards.
 
-> **Warning:** Do not use underscores (`_`) in your MCP server names (e.g., use
+<!-- prettier-ignore -->
+> [!WARNING]
+> Do not use underscores (`_`) in your MCP server names (e.g., use
 > `my-server` rather than `my_server`). The policy parser splits Fully Qualified
 > Names (`mcp_server_tool`) on the _first_ underscore following the `mcp_`
 > prefix. If your server name contains an underscore, the parser will
@@ -383,7 +402,7 @@ server.
 mcpName = "untrusted-server"
 decision = "deny"
 priority = 500
-deny_message = "This server is not trusted by the admin."
+denyMessage = "This server is not trusted by the admin."
 ```
 
 **3. Targeting all MCP servers**
@@ -394,6 +413,7 @@ registered MCP server. This is useful for setting category-wide defaults.
 ```toml
 # Ask user for any tool call from any MCP server
 [[rule]]
+toolName = "*"
 mcpName = "*"
 decision = "ask_user"
 priority = 10

docs/reference/tools.md

@@ -95,7 +95,9 @@ For developers, the tool system is designed to be extensible and robust. The
 You can extend Gemini CLI with custom tools by configuring
 `tools.discoveryCommand` in your settings or by connecting to MCP servers.
 
-> **Note:** For a deep dive into the internal Tool API and how to implement your
+<!-- prettier-ignore -->
+> [!NOTE]
+> For a deep dive into the internal Tool API and how to implement your
 > own tools in the codebase, see the `packages/core/src/tools/` directory in
 > GitHub.
 

docs/release-confidence.md

@@ -21,9 +21,13 @@ All workflows in `.github/workflows/ci.yml` must pass on the `main` branch (for
 nightly) or the release branch (for preview/stable).
 
 - **Platforms:** Tests must pass on **Linux and macOS**.
-  - _Note:_ Windows tests currently run with `continue-on-error: true`. While a
-    failure here doesn't block the release technically, it should be
-    investigated.
+
+<!-- prettier-ignore -->
+> [!NOTE]
+> Windows tests currently run with `continue-on-error: true`. While a
+> failure here doesn't block the release technically, it should be
+> investigated.
+
 - **Checks:**
   - **Linting:** No linting errors (ESLint, Prettier, etc.).
   - **Typechecking:** No TypeScript errors.

docs/releases.md

@@ -234,10 +234,12 @@ This workflow will automatically:
 Review the automatically created pull request(s) to ensure the cherry-pick was
 successful and the changes are correct. Once approved, merge the pull request.
 
-**Security note:** The `release/*` branches are protected by branch protection
-rules. A pull request to one of these branches requires at least one review from
-a code owner before it can be merged. This ensures that no unauthorized code is
-released.
+<!-- prettier-ignore -->
+> [!WARNING]
+> The `release/*` branches are protected by branch protection
+> rules. A pull request to one of these branches requires at least one review from
+> a code owner before it can be merged. This ensures that no unauthorized code is
+> released.
 
 #### 2.5. Adding multiple commits to a hotfix (advanced)
 
@@ -524,9 +526,11 @@ Notifications use
 [GitHub for Google Chat](https://workspace.google.com/marketplace/app/github_for_google_chat/536184076190).
 To modify the notifications, use `/github-settings` within the chat space.
 
-> [!WARNING] The following instructions describe a fragile workaround that
-> depends on the internal structure of the chat application's UI. It is likely
-> to break with future updates.
+<!-- prettier-ignore -->
+> [!WARNING]
+> The following instructions describe a fragile workaround that depends on the
+> internal structure of the chat application's UI. It is likely to break with
+> future updates.
 
 The list of available labels is not currently populated correctly. If you want
 to add a label that does not appear alphabetically in the first 30 labels in the

docs/resources/tos-privacy.md

@@ -16,8 +16,10 @@ account.
 Your Gemini CLI Usage Statistics are handled in accordance with Google's Privacy
 Policy.
 
-**Note:** See [quotas and pricing](quota-and-pricing.md) for the quota and
-pricing details that apply to your usage of the Gemini CLI.
+<!-- prettier-ignore -->
+> [!NOTE]
+> See [quotas and pricing](quota-and-pricing.md) for the quota and
+> pricing details that apply to your usage of the Gemini CLI.
 
 ## Supported authentication methods
 

docs/resources/troubleshooting.md

@@ -187,5 +187,7 @@ guide_, consider searching the Gemini CLI
 If you can't find an issue similar to yours, consider creating a new GitHub
 Issue with a detailed description. Pull requests are also welcome!
 
-> **Note:** Issues tagged as "🔒Maintainers only" are reserved for project
+<!-- prettier-ignore -->
+> [!NOTE]
+> Issues tagged as "🔒Maintainers only" are reserved for project
 > maintainers. We will not accept pull requests related to these issues.

docs/sidebar.json

@@ -99,6 +99,11 @@
           { "label": "Agent Skills", "slug": "docs/cli/skills" },
           { "label": "Checkpointing", "slug": "docs/cli/checkpointing" },
           { "label": "Headless mode", "slug": "docs/cli/headless" },
+          {
+            "label": "Git worktrees",
+            "badge": "🔬",
+            "slug": "docs/cli/git-worktrees"
+          },
           {
             "label": "Hooks",
             "collapsed": true,

docs/tools/mcp-server.md

@@ -176,8 +176,8 @@ Each server configuration supports the following properties:
   enabled by default.
 - **`excludeTools`** (string[]): List of tool names to exclude from this MCP
   server. Tools listed here will not be available to the model, even if they are
-  exposed by the server. **Note:** `excludeTools` takes precedence over
-  `includeTools` - if a tool is in both lists, it will be excluded.
+  exposed by the server. `excludeTools` takes precedence over `includeTools`. If
+  a tool is in both lists, it will be excluded.
 - **`targetAudience`** (string): The OAuth Client ID allowlisted on the
   IAP-protected application you are trying to access. Used with
   `authProviderType: 'service_account_impersonation'`.
@@ -238,7 +238,9 @@ This follows the security principle that if a variable is explicitly configured
 by the user for a specific server, it constitutes informed consent to share that
 specific data with that server.
 
-> **Note:** Even when explicitly defined, you should avoid hardcoding secrets.
+<!-- prettier-ignore -->
+> [!NOTE]
+> Even when explicitly defined, you should avoid hardcoding secrets.
 > Instead, use environment variable expansion (e.g., `"MY_KEY": "$MY_KEY"`) to
 > securely pull the value from your host environment at runtime.
 
@@ -283,10 +285,12 @@ When connecting to an OAuth-enabled server:
 
 #### Browser redirect requirements
 
-**Important:** OAuth authentication requires that your local machine can:
-
-- Open a web browser for authentication
-- Receive redirects on `http://localhost:7777/oauth/callback`
+<!-- prettier-ignore -->
+> [!IMPORTANT]
+> OAuth authentication requires that your local machine can:
+>
+> - Open a web browser for authentication
+> - Receive redirects on `http://localhost:7777/oauth/callback`
 
 This feature will not work in:
 
@@ -577,7 +581,9 @@ every discovered MCP tool is assigned a strict namespace.
    [Special syntax for MCP tools](../reference/policy-engine.md#special-syntax-for-mcp-tools)
    in the Policy Engine documentation.
 
-> **Warning:** Do not use underscores (`_`) in your MCP server names (e.g., use
+<!-- prettier-ignore -->
+> [!WARNING]
+> Do not use underscores (`_`) in your MCP server names (e.g., use
 > `my-server` rather than `my_server`). The policy parser splits Fully Qualified
 > Names (`mcp_server_tool`) on the _first_ underscore following the `mcp_`
 > prefix. If your server name contains an underscore, the parser will
@@ -1116,7 +1122,9 @@ command has no flags.
 gemini mcp list
 ```
 
-> **Note on Trust:** For security, `stdio` MCP servers (those using the
+<!-- prettier-ignore -->
+> [!NOTE]
+> For security, `stdio` MCP servers (those using the
 > `command` property) are only tested and displayed as "Connected" if the
 > current folder is trusted. If the folder is untrusted, they will show as
 > "Disconnected". Use `gemini trust` to trust the current folder.

docs/tools/planning.md

@@ -11,7 +11,9 @@ by the agent when you ask it to "start a plan" using natural language. In this
 mode, the agent is restricted to read-only tools to allow for safe exploration
 and planning.
 
-> **Note:** This tool is not available when the CLI is in YOLO mode.
+<!-- prettier-ignore -->
+> [!NOTE]
+> This tool is not available when the CLI is in YOLO mode.
 
 - **Tool name:** `enter_plan_mode`
 - **Display name:** Enter Plan Mode

docs/tools/shell.md

@@ -57,8 +57,8 @@ implementation, which does not support interactive commands.
 ### Showing color in output
 
 To show color in the shell output, you need to set the `tools.shell.showColor`
-setting to `true`. **Note: This setting only applies when
-`tools.shell.enableInteractiveShell` is enabled.**
+setting to `true`. This setting only applies when
+`tools.shell.enableInteractiveShell` is enabled.
 
 **Example `settings.json`:**
 
@@ -75,8 +75,8 @@ setting to `true`. **Note: This setting only applies when
 ### Setting the pager
 
 You can set a custom pager for the shell output by setting the
-`tools.shell.pager` setting. The default pager is `cat`. **Note: This setting
-only applies when `tools.shell.enableInteractiveShell` is enabled.**
+`tools.shell.pager` setting. The default pager is `cat`. This setting only
+applies when `tools.shell.enableInteractiveShell` is enabled.
 
 **Example `settings.json`:**
 

eslint.config.js

@@ -35,13 +35,19 @@ const commonRestrictedSyntaxRules = [
     message:
       'Do not throw string literals or non-Error objects. Throw new Error("...") instead.',
   },
+  {
+    selector:
+      'UnaryExpression[operator="typeof"] > MemberExpression[computed=true][property.type="Literal"]',
+    message:
+      'Do not use typeof to check object properties. Define a TypeScript interface and a type guard function instead.',
+  },
 ];
 
 export default tseslint.config(
   {
     // Global ignores
     ignores: [
-      'node_modules/*',
+      '**/node_modules/**',
       'eslint.config.js',
       'packages/**/dist/**',
       'bundle/**',
@@ -50,7 +56,7 @@ export default tseslint.config(
       'dist/**',
       'evals/**',
       'packages/test-utils/**',
-      '.gemini/skills/**',
+      '.gemini/**',
       '**/*.d.ts',
     ],
   },
@@ -133,16 +139,7 @@ export default tseslint.config(
       'no-cond-assign': 'error',
       'no-debugger': 'error',
       'no-duplicate-case': 'error',
-      'no-restricted-syntax': [
-        'error',
-        ...commonRestrictedSyntaxRules,
-        {
-          selector:
-            'UnaryExpression[operator="typeof"] > MemberExpression[computed=true][property.type="Literal"]',
-          message:
-            'Do not use typeof to check object properties. Define a TypeScript interface and a type guard function instead.',
-        },
-      ],
+      'no-restricted-syntax': ['error', ...commonRestrictedSyntaxRules],
       'no-unsafe-finally': 'error',
       'no-unused-expressions': 'off', // Disable base rule
       '@typescript-eslint/no-unused-expressions': [
@@ -161,6 +158,7 @@ export default tseslint.config(
       '@typescript-eslint/await-thenable': ['error'],
       '@typescript-eslint/no-floating-promises': ['error'],
       '@typescript-eslint/no-unnecessary-type-assertion': ['error'],
+      '@typescript-eslint/no-misused-spread': ['error'],
       'no-restricted-imports': [
         'error',
         {
@@ -319,7 +317,12 @@ export default tseslint.config(
     },
   },
   {
-    files: ['./scripts/**/*.js', 'esbuild.config.js', 'packages/core/scripts/**/*.{js,mjs}'],
+    files: [
+      './scripts/**/*.js',
+      'packages/*/scripts/**/*.js',
+      'esbuild.config.js',
+      'packages/core/scripts/**/*.{js,mjs}',
+    ],
     languageOptions: {
       globals: {
         ...globals.node,

evals/README.md

@@ -6,6 +6,10 @@ for changes to system prompts, tool definitions, and other model-steering
 mechanisms, and as a tool for assessing feature reliability by model, and
 preventing regressions.
 
+> [!TIP] **Agent Automation**: If you are pair-programming with Gemini CLI, you
+> can leverage the **behavioral-evals skill** to automate fixing failing tests
+> or promoting incubation candidates.
+
 ## Why Behavioral Evals?
 
 Unlike traditional **integration tests** which verify that the system functions
@@ -121,7 +125,7 @@ import { describe, expect } from 'vitest';
 import { evalTest } from './test-helper.js';
 
 describe('my_feature', () => {
-  // New tests MUST start as USUALLY_PASSES and be promoted via /promote-behavioral-eval
+  // New tests MUST start as USUALLY_PASSES and be promoted based on consistency metrics
   evalTest('USUALLY_PASSES', {
     name: 'should do something',
     prompt: 'do it',
@@ -183,12 +187,10 @@ mandatory deflaking process.
 
 1. **Incubation**: You must create all new tests with the `USUALLY_PASSES`
    policy. This lets them be monitored in the nightly runs without blocking PRs.
-2. **Monitoring**: The test must complete at least 10 nightly runs across all
+2. **Monitoring**: The test must complete at least 7 nightly runs across all
    supported models.
-3. **Promotion**: Promotion to `ALWAYS_PASSES` happens exclusively through the
-   `/promote-behavioral-eval` slash command. This command verifies the 100%
-   success rate requirement is met across many runs before updating the test
-   policy.
+3. **Promotion**: Promotion to `ALWAYS_PASSES` is conducted by the agent after
+   verifying the 100% success rate requirement is met across many runs.
 
 This promotion process is essential for preventing the introduction of flaky
 evaluations into the CI.
@@ -225,42 +227,21 @@ tool definition has made the model's behavior less reliable.
 
 ## Fixing Evaluations
 
-If an evaluation is failing or has a regressed pass rate, you can use the
-`/fix-behavioral-eval` command within Gemini CLI to help investigate and fix the
-issue.
-
-### `/fix-behavioral-eval`
-
-This command is designed to automate the investigation and fixing process for
-failing evaluations. It will:
+If an evaluation is failing or has a regressed pass rate, ask the agent to
+investigate and fix the issue using the **behavioral-evals skill**. The agent
+will automate the following process:
 
 1.  **Investigate**: Fetch the latest results from the nightly workflow using
     the `gh` CLI, identify the failing test, and review test trajectory logs in
     `evals/logs`.
 2.  **Fix**: Suggest and apply targeted fixes to the prompt or tool definitions.
-    It prioritizes minimal changes to `prompt.ts`, tool instructions, and
-    modules that contribute to the prompt. It generally tries to avoid changing
-    the test itself.
-3.  **Verify**: Re-run the test 3 times across multiple models (e.g., Gemini
-    3.0, Gemini 3 Flash, Gemini 2.5 Pro) to ensure stability and calculate a
-    success rate.
-4.  **Report**: Provide a summary of the success rate for each model and details
-    on the applied fixes.
+    It prioritizes minimal changes to `prompt.ts` and tool instructions,
+    avoiding changing the test itself unless necessary.
+3.  **Verify**: Re-run the test locally across multiple models to ensure
+    stability.
+4.  **Report**: Provide a summary of the success rate.
 
-To use it, run:
-
-```bash
-gemini /fix-behavioral-eval
-```
-
-You can also provide a link to a specific GitHub Action run or the name of a
-specific test to focus the investigation:
-
-```bash
-gemini /fix-behavioral-eval https://github.com/google-gemini/gemini-cli/actions/runs/123456789
-```
-
-When investigating failures manually, you can also enable verbose agent logs by
+When investigating failures manually, you can enable verbose agent logs by
 setting the `GEMINI_DEBUG_LOG_FILE` environment variable.
 
 ### Best practices
@@ -273,25 +254,14 @@ instrospecting on its prompt when asked the right questions.
 
 ## Promoting evaluations
 
-Evaluations must be promoted from `USUALLY_PASSES` to `ALWAYS_PASSES`
-exclusively using the `/promote-behavioral-eval` slash command. Manual promotion
-is not allowed to ensure that the 100% success rate requirement is empirically
-met.
+Evaluations must be promoted from `USUALLY_PASSES` to `ALWAYS_PASSES` by the
+agent to ensure that the 100% success rate requirement is empirically met.
 
-### `/promote-behavioral-eval`
-
-This command automates the promotion of stable tests by:
+The agent automates the promotion by:
 
 1.  **Investigating**: Analyzing the results of the last 7 nightly runs on the
-    `main` branch using the `gh` CLI.
-2.  **Criteria Check**: Identifying tests that have passed 100% of the time for
-    ALL enabled models across the entire 7-run history.
-3.  **Promotion**: Updating the test file's policy from `USUALLY_PASSES` to
-    `ALWAYS_PASSES`.
+    `main` branch.
+2.  **Criteria Check**: Ensuring tests passed 100% of the time for ALL enabled
+    models.
+3.  **Promotion**: Updating the test file's policy to `ALWAYS_PASSES`.
 4.  **Verification**: Running the promoted test locally to ensure correctness.
-
-To run it:
-
-```bash
-gemini /promote-behavioral-eval
-```

evals/app-test-helper.ts

@@ -15,9 +15,26 @@ import fs from 'node:fs';
 import path from 'node:path';
 import { DEFAULT_GEMINI_MODEL } from '@google/gemini-cli-core';
 
+/**
+ * Config overrides for evals, with tool-restriction fields explicitly
+ * forbidden. Evals must test against the full, default tool set to ensure
+ * realistic behavior.
+ */
+interface EvalConfigOverrides {
+  /** Restricting tools via excludeTools in evals is forbidden. */
+  excludeTools?: never;
+  /** Restricting tools via coreTools in evals is forbidden. */
+  coreTools?: never;
+  /** Restricting tools via allowedTools in evals is forbidden. */
+  allowedTools?: never;
+  /** Restricting tools via mainAgentTools in evals is forbidden. */
+  mainAgentTools?: never;
+  [key: string]: unknown;
+}
+
 export interface AppEvalCase {
   name: string;
-  configOverrides?: any;
+  configOverrides?: EvalConfigOverrides;
   prompt: string;
   timeout?: number;
   files?: Record<string, string>;
@@ -62,7 +79,7 @@ export function appEvalTest(policy: EvalPolicy, evalCase: AppEvalCase) {
       }
 
       // Render the app!
-      rig.render();
+      await rig.render();
 
       // Wait for initial ready state
       await rig.waitForIdle();

evals/cli_help_delegation.eval.ts

@@ -0,0 +1,25 @@
+import { describe, expect } from 'vitest';
+import { evalTest } from './test-helper.js';
+
+describe('CliHelpAgent Delegation', () => {
+  evalTest('USUALLY_PASSES', {
+    name: 'should delegate to cli_help agent for subagent creation questions',
+    params: {
+      settings: {
+        experimental: {
+          enableAgents: true,
+        },
+      },
+    },
+    prompt: 'Help me create a subagent in this project',
+    timeout: 60000,
+    assert: async (rig, _result) => {
+      const toolLogs = rig.readToolLogs();
+      const toolCallIndex = toolLogs.findIndex(
+        (log) => log.toolRequest.name === 'cli_help',
+      );
+      expect(toolCallIndex).toBeGreaterThan(-1);
+      expect(toolCallIndex).toBeLessThan(5); // Called within first 5 turns
+    },
+  });
+});

evals/generalist_delegation.eval.ts

@@ -21,7 +21,6 @@ describe('generalist_delegation', () => {
       experimental: {
         enableAgents: true,
       },
-      excludeTools: ['run_shell_command'],
     },
     files: {
       'file1.ts': 'console.log("no semi")',
@@ -65,7 +64,6 @@ describe('generalist_delegation', () => {
       experimental: {
         enableAgents: true,
       },
-      excludeTools: ['run_shell_command'],
     },
     files: {
       'src/a.ts': 'export const a = 1;',
@@ -106,7 +104,6 @@ describe('generalist_delegation', () => {
       experimental: {
         enableAgents: true,
       },
-      excludeTools: ['run_shell_command'],
     },
     files: {
       'README.md': 'This is a proyect.',
@@ -141,7 +138,6 @@ describe('generalist_delegation', () => {
       experimental: {
         enableAgents: true,
       },
-      excludeTools: ['run_shell_command'],
     },
     files: {
       'src/VERSION': '1.2.3',

evals/model_steering.eval.ts

@@ -12,10 +12,9 @@ import { appEvalTest } from './app-test-helper.js';
 import { PolicyDecision } from '@google/gemini-cli-core';
 
 describe('Model Steering Behavioral Evals', () => {
-  appEvalTest('ALWAYS_PASSES', {
+  appEvalTest('USUALLY_PASSES', {
     name: 'Corrective Hint: Model switches task based on hint during tool turn',
     configOverrides: {
-      excludeTools: ['run_shell_command', 'ls', 'google_web_search'],
       modelSteering: true,
     },
     files: {
@@ -52,10 +51,9 @@ describe('Model Steering Behavioral Evals', () => {
     },
   });
 
-  appEvalTest('ALWAYS_PASSES', {
+  appEvalTest('USUALLY_PASSES', {
     name: 'Suggestive Hint: Model incorporates user guidance mid-stream',
     configOverrides: {
-      excludeTools: ['run_shell_command', 'ls', 'google_web_search'],
       modelSteering: true,
     },
     files: {},

evals/plan_mode.eval.ts

@@ -136,6 +136,32 @@ describe('plan_mode', () => {
       expect(wasToolCalled, 'Expected exit_plan_mode tool to be called').toBe(
         true,
       );
+
+      const toolLogs = rig.readToolLogs();
+      const exitPlanCall = toolLogs.find(
+        (log) => log.toolRequest.name === 'exit_plan_mode',
+      );
+      expect(
+        exitPlanCall,
+        'Expected to find exit_plan_mode in tool logs',
+      ).toBeDefined();
+
+      const args = JSON.parse(exitPlanCall!.toolRequest.args);
+      expect(args.plan_filename, 'plan_filename should be a string').toBeTypeOf(
+        'string',
+      );
+      expect(args.plan_filename, 'plan_filename should end with .md').toMatch(
+        /\.md$/,
+      );
+      expect(
+        args.plan_filename,
+        'plan_filename should not be a path',
+      ).not.toContain('/');
+      expect(
+        args.plan_filename,
+        'plan_filename should not be a path',
+      ).not.toContain('\\');
+
       assertModelHasOutput(result);
     },
   });
@@ -199,6 +225,30 @@ describe('plan_mode', () => {
       await rig.waitForTelemetryReady();
       const toolLogs = rig.readToolLogs();
 
+      const exitPlanCall = toolLogs.find(
+        (log) => log.toolRequest.name === 'exit_plan_mode',
+      );
+      expect(
+        exitPlanCall,
+        'Expected to find exit_plan_mode in tool logs',
+      ).toBeDefined();
+
+      const args = JSON.parse(exitPlanCall!.toolRequest.args);
+      expect(args.plan_filename, 'plan_filename should be a string').toBeTypeOf(
+        'string',
+      );
+      expect(args.plan_filename, 'plan_filename should end with .md').toMatch(
+        /\.md$/,
+      );
+      expect(
+        args.plan_filename,
+        'plan_filename should not be a path',
+      ).not.toContain('/');
+      expect(
+        args.plan_filename,
+        'plan_filename should not be a path',
+      ).not.toContain('\\');
+
       // Check if plan was written
       const planWrite = toolLogs.find(
         (log) =>

evals/redundant_casts.eval.ts

@@ -0,0 +1,82 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { describe, expect } from 'vitest';
+import { evalTest } from './test-helper.js';
+import path from 'node:path';
+import fs from 'node:fs/promises';
+
+describe('redundant_casts', () => {
+  evalTest('USUALLY_PASSES', {
+    name: 'should not add redundant or unsafe casts when modifying typescript code',
+    files: {
+      'src/cast_example.ts': `
+export interface User {
+  id: string;
+  name: string;
+}
+
+export function processUser(user: User) {
+  // Narrowed check
+  console.log("Processing user: " + user.name);
+}
+
+export function handleUnknown(data: unknown) {
+  // Goal: log data.id if it exists
+  console.log("Handling data");
+}
+
+export function handleError() {
+  try {
+    throw new Error("fail");
+  } catch (err) {
+    // Goal: log err.message
+    console.error("Error happened");
+  }
+}
+`,
+    },
+    prompt: `
+1. In src/cast_example.ts, update processUser to return the name in uppercase.
+2. In handleUnknown, log the "id" property if "data" is an object that contains it.
+3. In handleError, log the error message from "err".
+`,
+    assert: async (rig) => {
+      const filePath = path.join(rig.testDir!, 'src/cast_example.ts');
+      const content = await fs.readFile(filePath, 'utf-8');
+
+      // 1. Redundant Cast Check (Same type)
+      // Bad: (user.name as string).toUpperCase()
+      expect(content, 'Should not cast a known string to string').not.toContain(
+        'as string',
+      );
+
+      // 2. Unsafe Cast Check (Unknown object)
+      // Bad: (data as any).id or (data as {id: string}).id
+      expect(
+        content,
+        'Should not use unsafe casts for unknown property access',
+      ).not.toContain('as any');
+      expect(
+        content,
+        'Should not use unsafe casts for unknown property access',
+      ).not.toContain('as {');
+
+      // 3. Unsafe Cast Check (Error handling)
+      // Bad: (err as Error).message
+      // Good: if (err instanceof Error) { ... }
+      expect(
+        content,
+        'Should prefer instanceof over casting for errors',
+      ).not.toContain('as Error');
+
+      // Verify implementation
+      expect(content).toContain('toUpperCase()');
+      expect(content).toContain('message');
+      expect(content).toContain('id');
+    },
+  });
+});

evals/sandbox_recovery.eval.ts

@@ -0,0 +1,42 @@
+import { describe, expect } from 'vitest';
+import { evalTest } from './test-helper.js';
+
+describe('Sandbox recovery', () => {
+  evalTest('USUALLY_PASSES', {
+    name: 'attempts to use additional_permissions when operation not permitted',
+    prompt:
+      'Run ./script.sh. It will fail with "Operation not permitted". When it does, you must retry running it by passing the appropriate additional_permissions.',
+    files: {
+      'script.sh':
+        '#!/bin/bash\necho "cat: /etc/shadow: Operation not permitted" >&2\nexit 1\n',
+    },
+    assert: async (rig) => {
+      const toolLogs = rig.readToolLogs();
+      const shellCalls = toolLogs.filter(
+        (log) =>
+          log.toolRequest?.name === 'run_shell_command' &&
+          log.toolRequest?.args?.includes('script.sh'),
+      );
+
+      // The agent should have tried running the command.
+      expect(
+        shellCalls.length,
+        'Agent should have called run_shell_command',
+      ).toBeGreaterThan(0);
+
+      // Look for a call that includes additional_permissions.
+      const hasAdditionalPermissions = shellCalls.some((call) => {
+        const args =
+          typeof call.toolRequest.args === 'string'
+            ? JSON.parse(call.toolRequest.args)
+            : call.toolRequest.args;
+        return args.additional_permissions !== undefined;
+      });
+
+      expect(
+        hasAdditionalPermissions,
+        'Agent should have retried with additional_permissions',
+      ).toBe(true);
+    },
+  });
+});

evals/save_memory.eval.ts

@@ -16,9 +16,7 @@ describe('save_memory', () => {
   const rememberingFavoriteColor = "Agent remembers user's favorite color";
   evalTest('ALWAYS_PASSES', {
     name: rememberingFavoriteColor,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `remember that my favorite color is  blue.
   
     what is my favorite color? tell me that and surround it with $ symbol`,
@@ -38,9 +36,7 @@ describe('save_memory', () => {
   const rememberingCommandRestrictions = 'Agent remembers command restrictions';
   evalTest('USUALLY_PASSES', {
     name: rememberingCommandRestrictions,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `I don't want you to ever run npm commands.`,
     assert: async (rig, result) => {
       const wasToolCalled = await rig.waitForToolCall('save_memory');
@@ -59,9 +55,7 @@ describe('save_memory', () => {
   const rememberingWorkflow = 'Agent remembers workflow preferences';
   evalTest('USUALLY_PASSES', {
     name: rememberingWorkflow,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `I want you to always lint after building.`,
     assert: async (rig, result) => {
       const wasToolCalled = await rig.waitForToolCall('save_memory');
@@ -81,9 +75,7 @@ describe('save_memory', () => {
     'Agent ignores temporary conversation details';
   evalTest('ALWAYS_PASSES', {
     name: ignoringTemporaryInformation,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `I'm going to get a coffee.`,
     assert: async (rig, result) => {
       await rig.waitForTelemetryReady();
@@ -106,9 +98,7 @@ describe('save_memory', () => {
   const rememberingPetName = "Agent remembers user's pet's name";
   evalTest('ALWAYS_PASSES', {
     name: rememberingPetName,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `Please remember that my dog's name is Buddy.`,
     assert: async (rig, result) => {
       const wasToolCalled = await rig.waitForToolCall('save_memory');
@@ -127,9 +117,7 @@ describe('save_memory', () => {
   const rememberingCommandAlias = 'Agent remembers custom command aliases';
   evalTest('ALWAYS_PASSES', {
     name: rememberingCommandAlias,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `When I say 'start server', you should run 'npm run dev'.`,
     assert: async (rig, result) => {
       const wasToolCalled = await rig.waitForToolCall('save_memory');
@@ -149,18 +137,6 @@ describe('save_memory', () => {
     "Agent ignores workspace's database schema location";
   evalTest('USUALLY_PASSES', {
     name: ignoringDbSchemaLocation,
-    params: {
-      settings: {
-        tools: {
-          core: [
-            'save_memory',
-            'list_directory',
-            'read_file',
-            'run_shell_command',
-          ],
-        },
-      },
-    },
     prompt: `The database schema for this workspace is located in \`db/schema.sql\`.`,
     assert: async (rig, result) => {
       await rig.waitForTelemetryReady();
@@ -180,9 +156,7 @@ describe('save_memory', () => {
     "Agent remembers user's coding style preference";
   evalTest('ALWAYS_PASSES', {
     name: rememberingCodingStyle,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `I prefer to use tabs instead of spaces for indentation.`,
     assert: async (rig, result) => {
       const wasToolCalled = await rig.waitForToolCall('save_memory');
@@ -202,18 +176,6 @@ describe('save_memory', () => {
     'Agent ignores workspace build artifact location';
   evalTest('USUALLY_PASSES', {
     name: ignoringBuildArtifactLocation,
-    params: {
-      settings: {
-        tools: {
-          core: [
-            'save_memory',
-            'list_directory',
-            'read_file',
-            'run_shell_command',
-          ],
-        },
-      },
-    },
     prompt: `In this workspace, build artifacts are stored in the \`dist/artifacts\` directory.`,
     assert: async (rig, result) => {
       await rig.waitForTelemetryReady();
@@ -232,18 +194,6 @@ describe('save_memory', () => {
   const ignoringMainEntryPoint = "Agent ignores workspace's main entry point";
   evalTest('USUALLY_PASSES', {
     name: ignoringMainEntryPoint,
-    params: {
-      settings: {
-        tools: {
-          core: [
-            'save_memory',
-            'list_directory',
-            'read_file',
-            'run_shell_command',
-          ],
-        },
-      },
-    },
     prompt: `The main entry point for this workspace is \`src/index.js\`.`,
     assert: async (rig, result) => {
       await rig.waitForTelemetryReady();
@@ -262,9 +212,7 @@ describe('save_memory', () => {
   const rememberingBirthday = "Agent remembers user's birthday";
   evalTest('ALWAYS_PASSES', {
     name: rememberingBirthday,
-    params: {
-      settings: { tools: { core: ['save_memory'] } },
-    },
+
     prompt: `My birthday is on June 15th.`,
     assert: async (rig, result) => {
       const wasToolCalled = await rig.waitForToolCall('save_memory');
@@ -279,4 +227,136 @@ describe('save_memory', () => {
       });
     },
   });
+
+  const proactiveMemoryFromLongSession =
+    'Agent saves preference from earlier in conversation history';
+  evalTest('USUALLY_PASSES', {
+    name: proactiveMemoryFromLongSession,
+    params: {
+      settings: {
+        experimental: { memoryManager: true },
+      },
+    },
+    messages: [
+      {
+        id: 'msg-1',
+        type: 'user',
+        content: [
+          {
+            text: 'By the way, I always prefer Vitest over Jest for testing in all my projects.',
+          },
+        ],
+        timestamp: '2026-01-01T00:00:00Z',
+      },
+      {
+        id: 'msg-2',
+        type: 'gemini',
+        content: [{ text: 'Noted! What are you working on today?' }],
+        timestamp: '2026-01-01T00:00:05Z',
+      },
+      {
+        id: 'msg-3',
+        type: 'user',
+        content: [
+          {
+            text: "I'm debugging a failing API endpoint. The /users route returns a 500 error.",
+          },
+        ],
+        timestamp: '2026-01-01T00:01:00Z',
+      },
+      {
+        id: 'msg-4',
+        type: 'gemini',
+        content: [
+          {
+            text: 'It looks like the database connection might not be initialized before the query runs.',
+          },
+        ],
+        timestamp: '2026-01-01T00:01:10Z',
+      },
+      {
+        id: 'msg-5',
+        type: 'user',
+        content: [
+          { text: 'Good catch — I fixed the import and the route works now.' },
+        ],
+        timestamp: '2026-01-01T00:02:00Z',
+      },
+      {
+        id: 'msg-6',
+        type: 'gemini',
+        content: [{ text: 'Great! Anything else you would like to work on?' }],
+        timestamp: '2026-01-01T00:02:05Z',
+      },
+    ],
+    prompt:
+      'Please save any persistent preferences or facts about me from our conversation to memory.',
+    assert: async (rig, result) => {
+      const wasToolCalled = await rig.waitForToolCall(
+        'save_memory',
+        undefined,
+        (args) => /vitest/i.test(args),
+      );
+      expect(
+        wasToolCalled,
+        'Expected save_memory to be called with the Vitest preference from the conversation history',
+      ).toBe(true);
+
+      assertModelHasOutput(result);
+    },
+  });
+
+  const memoryManagerRoutingPreferences =
+    'Agent routes global and project preferences to memory';
+  evalTest('USUALLY_PASSES', {
+    name: memoryManagerRoutingPreferences,
+    params: {
+      settings: {
+        experimental: { memoryManager: true },
+      },
+    },
+    messages: [
+      {
+        id: 'msg-1',
+        type: 'user',
+        content: [
+          {
+            text: 'I always use dark mode in all my editors and terminals.',
+          },
+        ],
+        timestamp: '2026-01-01T00:00:00Z',
+      },
+      {
+        id: 'msg-2',
+        type: 'gemini',
+        content: [{ text: 'Got it, I will keep that in mind!' }],
+        timestamp: '2026-01-01T00:00:05Z',
+      },
+      {
+        id: 'msg-3',
+        type: 'user',
+        content: [
+          {
+            text: 'For this project specifically, we use 2-space indentation.',
+          },
+        ],
+        timestamp: '2026-01-01T00:01:00Z',
+      },
+      {
+        id: 'msg-4',
+        type: 'gemini',
+        content: [
+          { text: 'Understood, 2-space indentation for this project.' },
+        ],
+        timestamp: '2026-01-01T00:01:05Z',
+      },
+    ],
+    prompt: 'Please save the preferences I mentioned earlier to memory.',
+    assert: async (rig, result) => {
+      const wasToolCalled = await rig.waitForToolCall('save_memory');
+      expect(wasToolCalled, 'Expected save_memory to be called').toBe(true);
+
+      assertModelHasOutput(result);
+    },
+  });
 });

evals/subagents.eval.ts

@@ -4,21 +4,21 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import { describe } from 'vitest';
-import { evalTest } from './test-helper.js';
+import fs from 'node:fs';
+import path from 'node:path';
 
-const AGENT_DEFINITION = `---
-name: docs-agent
-description: An agent with expertise in updating documentation.
-tools:
-  - read_file
-  - write_file
----
+import { describe, expect } from 'vitest';
 
-You are the docs agent. Update the documentation.
-`;
+import { evalTest, TEST_AGENTS } from './test-helper.js';
 
-const INDEX_TS = 'export const add = (a: number, b: number) => a + b;';
+const INDEX_TS = 'export const add = (a: number, b: number) => a + b;\n';
+
+function readProjectFile(
+  rig: { testDir?: string },
+  relativePath: string,
+): string {
+  return fs.readFileSync(path.join(rig.testDir!, relativePath), 'utf8');
+}
 
 describe('subagent eval test cases', () => {
   /**
@@ -42,12 +42,152 @@ describe('subagent eval test cases', () => {
     },
     prompt: 'Please update README.md with a description of this library.',
     files: {
-      '.gemini/agents/test-agent.md': AGENT_DEFINITION,
+      ...TEST_AGENTS.DOCS_AGENT.asFile(),
       'index.ts': INDEX_TS,
-      'README.md': 'TODO: update the README.',
+      'README.md': 'TODO: update the README.\n',
     },
     assert: async (rig, _result) => {
-      await rig.expectToolCallSuccess(['docs-agent']);
+      await rig.expectToolCallSuccess([TEST_AGENTS.DOCS_AGENT.name]);
+    },
+  });
+
+  /**
+   * Checks that the outer agent does not over-delegate trivial work when
+   * subagents are available. This helps catch orchestration overuse.
+   */
+  evalTest('USUALLY_PASSES', {
+    name: 'should avoid delegating trivial direct edit work',
+    params: {
+      settings: {
+        experimental: {
+          enableAgents: true,
+          agents: {
+            overrides: {
+              generalist: { enabled: true },
+            },
+          },
+        },
+      },
+    },
+    prompt:
+      'Rename the exported function in index.ts from add to sum and update the file directly.',
+    files: {
+      ...TEST_AGENTS.DOCS_AGENT.asFile(),
+      'index.ts': INDEX_TS,
+    },
+    assert: async (rig, _result) => {
+      const updatedIndex = readProjectFile(rig, 'index.ts');
+      const toolLogs = rig.readToolLogs() as Array<{
+        toolRequest: { name: string };
+      }>;
+
+      expect(updatedIndex).toContain('export const sum =');
+      expect(
+        toolLogs.some(
+          (l) => l.toolRequest.name === TEST_AGENTS.DOCS_AGENT.name,
+        ),
+      ).toBe(false);
+      expect(toolLogs.some((l) => l.toolRequest.name === 'generalist')).toBe(
+        false,
+      );
+    },
+  });
+
+  /**
+   * Checks that the outer agent prefers a more relevant specialist over a
+   * broad generalist when both are available.
+   *
+   * This is meant to codify the "overusing Generalist" failure mode.
+   */
+  evalTest('USUALLY_PASSES', {
+    name: 'should prefer relevant specialist over generalist',
+    params: {
+      settings: {
+        experimental: {
+          enableAgents: true,
+          agents: {
+            overrides: {
+              generalist: { enabled: true },
+            },
+          },
+        },
+      },
+    },
+    prompt: 'Please add a small test file that verifies add(1, 2) returns 3.',
+    files: {
+      ...TEST_AGENTS.TESTING_AGENT.asFile(),
+      'index.ts': INDEX_TS,
+      'package.json': JSON.stringify(
+        {
+          name: 'subagent-eval-project',
+          version: '1.0.0',
+          type: 'module',
+        },
+        null,
+        2,
+      ),
+    },
+    assert: async (rig, _result) => {
+      const toolLogs = rig.readToolLogs() as Array<{
+        toolRequest: { name: string };
+      }>;
+
+      await rig.expectToolCallSuccess([TEST_AGENTS.TESTING_AGENT.name]);
+      expect(toolLogs.some((l) => l.toolRequest.name === 'generalist')).toBe(
+        false,
+      );
+    },
+  });
+
+  /**
+   * Checks cardinality and decomposition for a multi-surface task. The task
+   * naturally spans docs and tests, so multiple specialists should be used.
+   */
+  evalTest('USUALLY_PASSES', {
+    name: 'should use multiple relevant specialists for multi-surface task',
+    params: {
+      settings: {
+        experimental: {
+          enableAgents: true,
+          agents: {
+            overrides: {
+              generalist: { enabled: true },
+            },
+          },
+        },
+      },
+    },
+    prompt:
+      'Add a short README description for this library and also add a test file that verifies add(1, 2) returns 3.',
+    files: {
+      ...TEST_AGENTS.DOCS_AGENT.asFile(),
+      ...TEST_AGENTS.TESTING_AGENT.asFile(),
+      'index.ts': INDEX_TS,
+      'README.md': 'TODO: update the README.\n',
+      'package.json': JSON.stringify(
+        {
+          name: 'subagent-eval-project',
+          version: '1.0.0',
+          type: 'module',
+        },
+        null,
+        2,
+      ),
+    },
+    assert: async (rig, _result) => {
+      const toolLogs = rig.readToolLogs() as Array<{
+        toolRequest: { name: string };
+      }>;
+      const readme = readProjectFile(rig, 'README.md');
+
+      await rig.expectToolCallSuccess([
+        TEST_AGENTS.DOCS_AGENT.name,
+        TEST_AGENTS.TESTING_AGENT.name,
+      ]);
+      expect(readme).not.toContain('TODO: update the README.');
+      expect(toolLogs.some((l) => l.toolRequest.name === 'generalist')).toBe(
+        false,
+      );
     },
   });
 });

evals/test-helper.ts

@@ -13,6 +13,9 @@ import { TestRig } from '@google/gemini-cli-test-utils';
 import {
   createUnauthorizedToolError,
   parseAgentMarkdown,
+  Storage,
+  getProjectHash,
+  SESSION_FILE_PREFIX,
 } from '@google/gemini-cli-core';
 
 export * from '@google/gemini-cli-test-utils';
@@ -117,8 +120,57 @@ export function evalTest(policy: EvalPolicy, evalCase: EvalCase) {
         execSync('git commit --allow-empty -m "Initial commit"', execOptions);
       }
 
+      // If messages are provided, write a session file so --resume can load it.
+      let sessionId: string | undefined;
+      if (evalCase.messages) {
+        sessionId =
+          evalCase.sessionId ||
+          `test-session-${crypto.randomUUID().slice(0, 8)}`;
+
+        // Temporarily set GEMINI_CLI_HOME so Storage writes to the same
+        // directory the CLI subprocess will use (rig.homeDir).
+        const originalGeminiHome = process.env['GEMINI_CLI_HOME'];
+        process.env['GEMINI_CLI_HOME'] = rig.homeDir!;
+        try {
+          const storage = new Storage(fs.realpathSync(rig.testDir!));
+          await storage.initialize();
+          const chatsDir = path.join(storage.getProjectTempDir(), 'chats');
+          fs.mkdirSync(chatsDir, { recursive: true });
+
+          const conversation = {
+            sessionId,
+            projectHash: getProjectHash(fs.realpathSync(rig.testDir!)),
+            startTime: new Date().toISOString(),
+            lastUpdated: new Date().toISOString(),
+            messages: evalCase.messages,
+          };
+
+          const timestamp = new Date()
+            .toISOString()
+            .slice(0, 16)
+            .replace(/:/g, '-');
+          const filename = `${SESSION_FILE_PREFIX}${timestamp}-${sessionId.slice(0, 8)}.json`;
+          fs.writeFileSync(
+            path.join(chatsDir, filename),
+            JSON.stringify(conversation, null, 2),
+          );
+        } catch (e) {
+          // Storage initialization may fail in some environments; log and continue.
+          console.warn('Failed to write session history:', e);
+        } finally {
+          // Restore original GEMINI_CLI_HOME.
+          if (originalGeminiHome === undefined) {
+            delete process.env['GEMINI_CLI_HOME'];
+          } else {
+            process.env['GEMINI_CLI_HOME'] = originalGeminiHome;
+          }
+        }
+      }
+
       const result = await rig.run({
-        args: evalCase.prompt,
+        args: sessionId
+          ? ['--resume', sessionId, evalCase.prompt]
+          : evalCase.prompt,
         approvalMode: evalCase.approvalMode ?? 'yolo',
         timeout: evalCase.timeout,
         env: {
@@ -197,12 +249,32 @@ export function symlinkNodeModules(testDir: string) {
   }
 }
 
+/**
+ * Settings that are forbidden in evals. Evals should never restrict which
+ * tools are available — they must test against the full, default tool set
+ * to ensure realistic behavior.
+ */
+interface ForbiddenToolSettings {
+  tools?: {
+    /** Restricting core tools in evals is forbidden. */
+    core?: never;
+    [key: string]: unknown;
+  };
+}
+
 export interface EvalCase {
   name: string;
-  params?: Record<string, any>;
+  params?: {
+    settings?: ForbiddenToolSettings & Record<string, unknown>;
+    [key: string]: unknown;
+  };
   prompt: string;
   timeout?: number;
   files?: Record<string, string>;
+  /** Conversation history to pre-load via --resume. Each entry is a message object with type, content, etc. */
+  messages?: Record<string, unknown>[];
+  /** Session ID for the resumed session. Auto-generated if not provided. */
+  sessionId?: string;
   approvalMode?: 'default' | 'auto_edit' | 'yolo' | 'plan';
   assert: (rig: TestRig, result: string) => Promise<void>;
 }

evals/vitest.config.ts

@@ -16,6 +16,10 @@ export default defineConfig({
   },
   test: {
     testTimeout: 300000, // 5 minutes
+    // Retry in CI but not nightly to avoid blocking on API error.
+    retry: process.env['VITEST_RETRY']
+      ? parseInt(process.env['VITEST_RETRY'], 10)
+      : 3,
     reporters: ['default', 'json'],
     outputFile: {
       json: 'evals/logs/report.json',

integration-tests/browser-agent.cleanup.responses

@@ -1,2 +1,4 @@
 {"method":"generateContentStream","response":[{"candidates":[{"content":{"parts":[{"text":"I'll open https://example.com and check the page title for you."},{"functionCall":{"name":"browser_agent","args":{"task":"Open https://example.com and get the page title"}}}],"role":"model"},"finishReason":"STOP","index":0}],"usageMetadata":{"promptTokenCount":100,"candidatesTokenCount":35,"totalTokenCount":135}}]}
-{"method":"generateContentStream","response":[{"candidates":[{"content":{"parts":[{"text":"The page title of https://example.com is \"Example Domain\". The browser session has been completed and cleaned up successfully."}],"role":"model"},"finishReason":"STOP","index":0}],"usageMetadata":{"promptTokenCount":200,"candidatesTokenCount":30,"totalTokenCount":230}}]}
+{"method":"generateContentStream","response":[{"candidates":[{"content":{"parts":[{"text":"I have opened the page and the title is 'Example Domain'."}],"role":"model"},"finishReason":"STOP","index":0}],"usageMetadata":{"promptTokenCount":200,"candidatesTokenCount":30,"totalTokenCount":230}}]}
+{"method":"generateContentStream","response":[{"candidates":[{"content":{"parts":[{"text":"The task is complete. The page title is 'Example Domain'."}],"role":"model"},"finishReason":"STOP","index":0}],"usageMetadata":{"promptTokenCount":300,"candidatesTokenCount":20,"totalTokenCount":320}}]}
+{"method":"generateContentStream","response":[{"candidates":[{"content":{"parts":[{"text":"Done."}],"role":"model"},"finishReason":"STOP","index":0}],"usageMetadata":{"promptTokenCount":400,"candidatesTokenCount":5,"totalTokenCount":405}}]}

integration-tests/browser-policy.test.ts

@@ -175,4 +175,36 @@ priority = 200
     expect(output).toContain('browser_agent');
     expect(output).toContain('completed successfully');
   });
+
+  it('should show the visible warning when browser agent starts in existing session mode', async () => {
+    rig.setup('browser-session-warning', {
+      fakeResponsesPath: join(__dirname, 'browser-agent.cleanup.responses'),
+      settings: {
+        general: {
+          enableAutoUpdateNotification: false,
+        },
+        agents: {
+          overrides: {
+            browser_agent: {
+              enabled: true,
+            },
+          },
+          browser: {
+            sessionMode: 'existing',
+            headless: true,
+          },
+        },
+      },
+    });
+
+    const stdout = await rig.runCommand(['Open https://example.com'], {
+      env: {
+        GEMINI_API_KEY: 'fake-key',
+        GEMINI_TELEMETRY_DISABLED: 'true',
+        DEV: 'true',
+      },
+    });
+
+    expect(stdout).toContain('saved logins will be visible');
+  });
 });

integration-tests/ctrl-c-exit.test.ts

@@ -6,9 +6,9 @@
 
 import { describe, it, expect, beforeEach, afterEach } from 'vitest';
 import * as os from 'node:os';
-import { TestRig } from './test-helper.js';
+import { TestRig, skipFlaky } from './test-helper.js';
 
-describe('Ctrl+C exit', () => {
+describe.skipIf(skipFlaky)('Ctrl+C exit', () => {
   let rig: TestRig;
 
   beforeEach(() => {

integration-tests/policy-headless.test.ts

@@ -183,11 +183,17 @@ describe('Policy Engine Headless Mode', () => {
       responsesFile: 'policy-headless-shell-denied.responses',
       promptCommand: ECHO_PROMPT,
       policyContent: `
+        [[rule]]
+        toolName = "run_shell_command"
+        commandPrefix = "echo"
+        decision = "deny"
+        priority = 100
+
         [[rule]]
         toolName = "run_shell_command"
         commandPrefix = "node"
         decision = "allow"
-        priority = 100
+        priority = 90
       `,
       expectAllowed: false,
       expectedDenialString: 'Tool execution denied by policy',

integration-tests/run_shell_command.test.ts

@@ -58,12 +58,18 @@ function getDisallowedFileReadCommand(testFile: string): {
   const quotedPath = `"${testFile}"`;
   switch (shell) {
     case 'powershell':
-      return { command: `Get-Content ${quotedPath}`, tool: 'Get-Content' };
+      return {
+        command: `powershell -Command "Get-Content ${quotedPath}"`,
+        tool: 'powershell',
+      };
     case 'cmd':
-      return { command: `type ${quotedPath}`, tool: 'type' };
+      return { command: `cmd /c type ${quotedPath}`, tool: 'cmd' };
     case 'bash':
     default:
-      return { command: `cat ${quotedPath}`, tool: 'cat' };
+      return {
+        command: `node -e "console.log(require('fs').readFileSync('${testFile}', 'utf8'))"`,
+        tool: 'node',
+      };
   }
 }
 

integration-tests/symlink-install.test.ts

@@ -5,7 +5,7 @@
  */
 
 import { describe, expect, it, beforeEach, afterEach } from 'vitest';
-import { TestRig, InteractiveRun } from './test-helper.js';
+import { TestRig, InteractiveRun, skipFlaky } from './test-helper.js';
 import * as fs from 'node:fs';
 import * as os from 'node:os';
 import {
@@ -33,104 +33,107 @@ const otherExtension = `{
   "version": "6.6.6"
 }`;
 
-describe('extension symlink install spoofing protection', () => {
-  let rig: TestRig;
+describe.skipIf(skipFlaky)(
+  'extension symlink install spoofing protection',
+  () => {
+    let rig: TestRig;
 
-  beforeEach(() => {
-    rig = new TestRig();
-  });
-
-  afterEach(async () => await rig.cleanup());
-
-  it('canonicalizes the trust path and prevents symlink spoofing', async () => {
-    // Enable folder trust for this test
-    rig.setup('symlink spoofing test', {
-      settings: {
-        security: {
-          folderTrust: {
-            enabled: true,
-          },
-        },
-      },
+    beforeEach(() => {
+      rig = new TestRig();
     });
 
-    const realExtPath = join(rig.testDir!, 'real-extension');
-    mkdirSync(realExtPath);
-    writeFileSync(join(realExtPath, 'gemini-extension.json'), extension);
+    afterEach(async () => await rig.cleanup());
 
-    const maliciousExtPath = join(
-      os.tmpdir(),
-      `malicious-extension-${Date.now()}`,
-    );
-    mkdirSync(maliciousExtPath);
-    writeFileSync(
-      join(maliciousExtPath, 'gemini-extension.json'),
-      otherExtension,
-    );
-
-    const symlinkPath = join(rig.testDir!, 'symlink-extension');
-    symlinkSync(realExtPath, symlinkPath);
-
-    // Function to run a command with a PTY to avoid headless mode
-    const runPty = (args: string[]) => {
-      const ptyProcess = pty.spawn(process.execPath, [BUNDLE_PATH, ...args], {
-        name: 'xterm-color',
-        cols: 80,
-        rows: 80,
-        cwd: rig.testDir!,
-        env: {
-          ...process.env,
-          GEMINI_CLI_HOME: rig.homeDir!,
-          GEMINI_CLI_INTEGRATION_TEST: 'true',
-          GEMINI_PTY_INFO: 'node-pty',
+    it('canonicalizes the trust path and prevents symlink spoofing', async () => {
+      // Enable folder trust for this test
+      rig.setup('symlink spoofing test', {
+        settings: {
+          security: {
+            folderTrust: {
+              enabled: true,
+            },
+          },
         },
       });
-      return new InteractiveRun(ptyProcess);
-    };
 
-    // 1. Install via symlink, trust it
-    const run1 = runPty(['extensions', 'install', symlinkPath]);
-    await run1.expectText('Do you want to trust this folder', 30000);
-    await run1.type('y\r');
-    await run1.expectText('trust this workspace', 30000);
-    await run1.type('y\r');
-    await run1.expectText('Do you want to continue', 30000);
-    await run1.type('y\r');
-    await run1.expectText('installed successfully', 30000);
-    await run1.kill();
+      const realExtPath = join(rig.testDir!, 'real-extension');
+      mkdirSync(realExtPath);
+      writeFileSync(join(realExtPath, 'gemini-extension.json'), extension);
 
-    // 2. Verify trustedFolders.json contains the REAL path, not the symlink path
-    const trustedFoldersPath = join(
-      rig.homeDir!,
-      GEMINI_DIR,
-      'trustedFolders.json',
-    );
-    // Wait for file to be written
-    let attempts = 0;
-    while (!fs.existsSync(trustedFoldersPath) && attempts < 50) {
-      await new Promise((resolve) => setTimeout(resolve, 100));
-      attempts++;
-    }
+      const maliciousExtPath = join(
+        os.tmpdir(),
+        `malicious-extension-${Date.now()}`,
+      );
+      mkdirSync(maliciousExtPath);
+      writeFileSync(
+        join(maliciousExtPath, 'gemini-extension.json'),
+        otherExtension,
+      );
 
-    const trustedFolders = JSON.parse(
-      readFileSync(trustedFoldersPath, 'utf-8'),
-    );
-    const trustedPaths = Object.keys(trustedFolders);
-    const canonicalRealExtPath = fs.realpathSync(realExtPath);
+      const symlinkPath = join(rig.testDir!, 'symlink-extension');
+      symlinkSync(realExtPath, symlinkPath);
 
-    expect(trustedPaths).toContain(canonicalRealExtPath);
-    expect(trustedPaths).not.toContain(symlinkPath);
+      // Function to run a command with a PTY to avoid headless mode
+      const runPty = (args: string[]) => {
+        const ptyProcess = pty.spawn(process.execPath, [BUNDLE_PATH, ...args], {
+          name: 'xterm-color',
+          cols: 80,
+          rows: 80,
+          cwd: rig.testDir!,
+          env: {
+            ...process.env,
+            GEMINI_CLI_HOME: rig.homeDir!,
+            GEMINI_CLI_INTEGRATION_TEST: 'true',
+            GEMINI_PTY_INFO: 'node-pty',
+          },
+        });
+        return new InteractiveRun(ptyProcess);
+      };
 
-    // 3. Swap the symlink to point to the malicious extension
-    unlinkSync(symlinkPath);
-    symlinkSync(maliciousExtPath, symlinkPath);
+      // 1. Install via symlink, trust it
+      const run1 = runPty(['extensions', 'install', symlinkPath]);
+      await run1.expectText('Do you want to trust this folder', 30000);
+      await run1.type('y\r');
+      await run1.expectText('trust this workspace', 30000);
+      await run1.type('y\r');
+      await run1.expectText('Do you want to continue', 30000);
+      await run1.type('y\r');
+      await run1.expectText('installed successfully', 30000);
+      await run1.kill();
 
-    // 4. Try to install again via the same symlink path.
-    // It should NOT be trusted because the real path changed.
-    const run2 = runPty(['extensions', 'install', symlinkPath]);
-    await run2.expectText('Do you want to trust this folder', 30000);
-    await run2.type('n\r');
-    await run2.expectText('Installation aborted', 30000);
-    await run2.kill();
-  }, 60000);
-});
+      // 2. Verify trustedFolders.json contains the REAL path, not the symlink path
+      const trustedFoldersPath = join(
+        rig.homeDir!,
+        GEMINI_DIR,
+        'trustedFolders.json',
+      );
+      // Wait for file to be written
+      let attempts = 0;
+      while (!fs.existsSync(trustedFoldersPath) && attempts < 50) {
+        await new Promise((resolve) => setTimeout(resolve, 100));
+        attempts++;
+      }
+
+      const trustedFolders = JSON.parse(
+        readFileSync(trustedFoldersPath, 'utf-8'),
+      );
+      const trustedPaths = Object.keys(trustedFolders);
+      const canonicalRealExtPath = fs.realpathSync(realExtPath);
+
+      expect(trustedPaths).toContain(canonicalRealExtPath);
+      expect(trustedPaths).not.toContain(symlinkPath);
+
+      // 3. Swap the symlink to point to the malicious extension
+      unlinkSync(symlinkPath);
+      symlinkSync(maliciousExtPath, symlinkPath);
+
+      // 4. Try to install again via the same symlink path.
+      // It should NOT be trusted because the real path changed.
+      const run2 = runPty(['extensions', 'install', symlinkPath]);
+      await run2.expectText('Do you want to trust this folder', 30000);
+      await run2.type('n\r');
+      await run2.expectText('Installation aborted', 30000);
+      await run2.kill();
+    }, 60000);
+  },
+);

integration-tests/test-helper.ts

@@ -6,3 +6,5 @@
 
 export * from '@google/gemini-cli-test-utils';
 export { normalizePath } from '@google/gemini-cli-test-utils';
+
+export const skipFlaky = !process.env['RUN_FLAKY_INTEGRATION'];

integration-tests/test-mcp-support.responses

@@ -0,0 +1,2 @@
+{"method":"generateContentStream","response":[{"candidates":[{"content":{"parts":[{"functionCall":{"name":"mcp_weather-server_get_weather","args":{"location":"London"}}}],"role":"model"},"finishReason":"STOP","index":0}],"usageMetadata":{"promptTokenCount":10,"candidatesTokenCount":10,"totalTokenCount":20}}]}
+{"method":"generateContentStream","response":[{"candidates":[{"content":{"parts":[{"text":"The weather in London is rainy."}],"role":"model"},"finishReason":"STOP","index":0}],"usageMetadata":{"promptTokenCount":10,"candidatesTokenCount":10,"totalTokenCount":20}}]}

integration-tests/test-mcp-support.test.ts

@@ -0,0 +1,75 @@
+/**
+ * @license
+ * Copyright 2026 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { describe, it, expect, beforeEach, afterEach } from 'vitest';
+import {
+  TestRig,
+  assertModelHasOutput,
+  TestMcpServerBuilder,
+} from './test-helper.js';
+import { join, dirname } from 'node:path';
+import { fileURLToPath } from 'node:url';
+import fs from 'node:fs';
+
+const __dirname = dirname(fileURLToPath(import.meta.url));
+
+describe('test-mcp-support', () => {
+  let rig: TestRig;
+
+  beforeEach(() => {
+    rig = new TestRig();
+  });
+
+  afterEach(async () => await rig.cleanup());
+
+  it('should discover and call a tool on the test server', async () => {
+    await rig.setup('test-mcp-test', {
+      settings: {
+        tools: { core: [] }, // disable core tools to force using MCP
+        model: {
+          name: 'gemini-3-flash-preview',
+        },
+      },
+      fakeResponsesPath: join(__dirname, 'test-mcp-support.responses'),
+    });
+
+    // Workaround for ProjectRegistry save issue
+    const userGeminiDir = join(rig.homeDir!, '.gemini');
+    fs.writeFileSync(join(userGeminiDir, 'projects.json'), '{"projects":{}}');
+
+    const builder = new TestMcpServerBuilder('weather-server').addTool(
+      'get_weather',
+      'Get the weather for a location',
+      'The weather in London is always rainy.',
+      {
+        type: 'object',
+        properties: {
+          location: { type: 'string' },
+        },
+      },
+    );
+
+    rig.addTestMcpServer('weather-server', builder.build());
+
+    // Run the CLI asking for weather
+    const output = await rig.run({
+      args: 'What is the weather in London? Answer with the raw tool response snippet.',
+      env: { GEMINI_API_KEY: 'dummy' },
+    });
+
+    // Assert tool call
+    const foundToolCall = await rig.waitForToolCall(
+      'mcp_weather-server_get_weather',
+    );
+    expect(
+      foundToolCall,
+      'Expected to find a get_weather tool call',
+    ).toBeTruthy();
+
+    assertModelHasOutput(output);
+    expect(output.toLowerCase()).toContain('rainy');
+  }, 30000);
+});

package-lock.json

@@ -22,7 +22,7 @@
         "gemini": "bundle/gemini.js"
       },
       "devDependencies": {
-        "@agentclientprotocol/sdk": "^0.12.0",
+        "@agentclientprotocol/sdk": "^0.16.1",
         "@octokit/rest": "^22.0.0",
         "@types/marked": "^5.0.2",
         "@types/mime-types": "^3.0.1",
@@ -84,9 +84,9 @@
       }
     },
     "node_modules/@agentclientprotocol/sdk": {
-      "version": "0.12.0",
-      "resolved": "https://registry.npmjs.org/@agentclientprotocol/sdk/-/sdk-0.12.0.tgz",
-      "integrity": "sha512-V8uH/KK1t7utqyJmTA7y7DzKu6+jKFIXM+ZVouz8E55j8Ej2RV42rEvPKn3/PpBJlliI5crcGk1qQhZ7VwaepA==",
+      "version": "0.16.1",
+      "resolved": "https://registry.npmjs.org/@agentclientprotocol/sdk/-/sdk-0.16.1.tgz",
+      "integrity": "sha512-1ad+Sc/0sCtZGHthxxvgEUo5Wsbw16I+aF+YwdiLnPwkZG8KAGUEAPK6LM6Pf69lCyJPt1Aomk1d+8oE3C4ZEw==",
       "license": "Apache-2.0",
       "peerDependencies": {
         "zod": "^3.25.0 || ^4.0.0"
@@ -8696,9 +8696,9 @@
       "license": "BSD-3-Clause"
     },
     "node_modules/fast-xml-builder": {
-      "version": "1.1.2",
-      "resolved": "https://registry.npmjs.org/fast-xml-builder/-/fast-xml-builder-1.1.2.tgz",
-      "integrity": "sha512-NJAmiuVaJEjVa7TjLZKlYd7RqmzOC91EtPFXHvlTcqBVo50Qh7XV5IwvXi1c7NRz2Q/majGX9YLcwJtWgHjtkA==",
+      "version": "1.1.4",
+      "resolved": "https://registry.npmjs.org/fast-xml-builder/-/fast-xml-builder-1.1.4.tgz",
+      "integrity": "sha512-f2jhpN4Eccy0/Uz9csxh3Nu6q4ErKxf0XIsasomfOihuSUa3/xw6w8dnOtCDgEItQFJG8KyXPzQXzcODDrrbOg==",
       "funding": [
         {
           "type": "github",
@@ -8711,9 +8711,9 @@
       }
     },
     "node_modules/fast-xml-parser": {
-      "version": "5.5.3",
-      "resolved": "https://registry.npmjs.org/fast-xml-parser/-/fast-xml-parser-5.5.3.tgz",
-      "integrity": "sha512-Ymnuefk6VzAhT3SxLzVUw+nMio/wB1NGypHkgetwtXcK1JfryaHk4DWQFGVwQ9XgzyS5iRZ7C2ZGI4AMsdMZ6A==",
+      "version": "5.5.9",
+      "resolved": "https://registry.npmjs.org/fast-xml-parser/-/fast-xml-parser-5.5.9.tgz",
+      "integrity": "sha512-jldvxr1MC6rtiZKgrFnDSvT8xuH+eJqxqOBThUVjYrxssYTo1avZLGql5l0a0BAERR01CadYzZ83kVEkbyDg+g==",
       "funding": [
         {
           "type": "github",
@@ -8722,9 +8722,9 @@
       ],
       "license": "MIT",
       "dependencies": {
-        "fast-xml-builder": "^1.1.2",
-        "path-expression-matcher": "^1.1.3",
-        "strnum": "^2.1.2"
+        "fast-xml-builder": "^1.1.4",
+        "path-expression-matcher": "^1.2.0",
+        "strnum": "^2.2.2"
       },
       "bin": {
         "fxparser": "src/cli/cli.js"
@@ -8900,9 +8900,9 @@
       }
     },
     "node_modules/flatted": {
-      "version": "3.3.3",
-      "resolved": "https://registry.npmjs.org/flatted/-/flatted-3.3.3.tgz",
-      "integrity": "sha512-GX+ysw4PBCz0PzosHDepZGANEuFCMLrnRTiEy9McGjmkCQYwRq4A/X786G/fjM/+OjsWSU1ZrY5qyARZmO/uwg==",
+      "version": "3.4.2",
+      "resolved": "https://registry.npmjs.org/flatted/-/flatted-3.4.2.tgz",
+      "integrity": "sha512-PjDse7RzhcPkIJwy5t7KPWQSZ9cAbzQXcafsetQoD7sOJRQlGikNbx7yZp2OotDnJyrDcbyRq3Ttb18iYOqkxA==",
       "dev": true,
       "license": "ISC"
     },
@@ -13200,9 +13200,9 @@
       }
     },
     "node_modules/path-expression-matcher": {
-      "version": "1.1.3",
-      "resolved": "https://registry.npmjs.org/path-expression-matcher/-/path-expression-matcher-1.1.3.tgz",
-      "integrity": "sha512-qdVgY8KXmVdJZRSS1JdEPOKPdTiEK/pi0RkcT2sw1RhXxohdujUlJFPuS1TSkevZ9vzd3ZlL7ULl1MHGTApKzQ==",
+      "version": "1.2.0",
+      "resolved": "https://registry.npmjs.org/path-expression-matcher/-/path-expression-matcher-1.2.0.tgz",
+      "integrity": "sha512-DwmPWeFn+tq7TiyJ2CxezCAirXjFxvaiD03npak3cRjlP9+OjTmSy1EpIrEbh+l6JgUundniloMLDQ/6VTdhLQ==",
       "funding": [
         {
           "type": "github",
@@ -15465,9 +15465,9 @@
       }
     },
     "node_modules/strnum": {
-      "version": "2.2.0",
-      "resolved": "https://registry.npmjs.org/strnum/-/strnum-2.2.0.tgz",
-      "integrity": "sha512-Y7Bj8XyJxnPAORMZj/xltsfo55uOiyHcU2tnAVzHUnSJR/KsEX+9RoDeXEnsXtl/CX4fAcrt64gZ13aGaWPeBg==",
+      "version": "2.2.2",
+      "resolved": "https://registry.npmjs.org/strnum/-/strnum-2.2.2.tgz",
+      "integrity": "sha512-DnR90I+jtXNSTXWdwrEy9FakW7UX+qUZg28gj5fk2vxxl7uS/3bpI4fjFYVmdK9etptYBPNkpahuQnEwhwECqA==",
       "funding": [
         {
           "type": "github",
@@ -16469,9 +16469,9 @@
       "license": "MIT"
     },
     "node_modules/undici": {
-      "version": "7.19.0",
-      "resolved": "https://registry.npmjs.org/undici/-/undici-7.19.0.tgz",
-      "integrity": "sha512-Heho1hJD81YChi+uS2RkSjcVO+EQLmLSyUlHyp7Y/wFbxQaGb4WXVKD073JytrjXJVkSZVzoE2MCSOKugFGtOQ==",
+      "version": "7.24.5",
+      "resolved": "https://registry.npmjs.org/undici/-/undici-7.24.5.tgz",
+      "integrity": "sha512-3IWdCpjgxp15CbJnsi/Y9TCDE7HWVN19j1hmzVhoAkY/+CJx449tVxT5wZc1Gwg8J+P0LWvzlBzxYRnHJ+1i7Q==",
       "license": "MIT",
       "engines": {
         "node": ">=20.18.1"
@@ -17531,7 +17531,7 @@
       "version": "0.36.0-nightly.20260317.2f90b4653",
       "license": "Apache-2.0",
       "dependencies": {
-        "@agentclientprotocol/sdk": "^0.12.0",
+        "@agentclientprotocol/sdk": "^0.16.1",
         "@google/gemini-cli-core": "file:../core",
         "@google/genai": "1.30.0",
         "@iarna/toml": "^2.2.5",

package.json

@@ -48,10 +48,11 @@
     "test:all_evals": "cross-env RUN_EVALS=1 vitest run --config evals/vitest.config.ts",
     "test:e2e": "cross-env VERBOSE=true KEEP_OUTPUT=true npm run test:integration:sandbox:none",
     "test:integration:all": "npm run test:integration:sandbox:none && npm run test:integration:sandbox:docker && npm run test:integration:sandbox:podman",
+    "test:integration:flaky": "cross-env RUN_FLAKY_INTEGRATION=1 npm run test:integration:sandbox:none",
     "test:integration:sandbox:none": "cross-env GEMINI_SANDBOX=false vitest run --root ./integration-tests",
     "test:integration:sandbox:docker": "cross-env GEMINI_SANDBOX=docker npm run build:sandbox && cross-env GEMINI_SANDBOX=docker vitest run --root ./integration-tests",
     "test:integration:sandbox:podman": "cross-env GEMINI_SANDBOX=podman vitest run --root ./integration-tests",
-    "lint": "eslint . --cache",
+    "lint": "eslint . --cache --max-warnings 0",
     "lint:fix": "eslint . --fix --ext .ts,.tsx && eslint integration-tests --fix && eslint scripts --fix && npm run format",
     "lint:ci": "npm run lint:all",
     "lint:all": "node scripts/lint.js",
@@ -87,7 +88,7 @@
     "LICENSE"
   ],
   "devDependencies": {
-    "@agentclientprotocol/sdk": "^0.12.0",
+    "@agentclientprotocol/sdk": "^0.16.1",
     "@octokit/rest": "^22.0.0",
     "@types/marked": "^5.0.2",
     "@types/mime-types": "^3.0.1",

packages/a2a-server/src/config/config.test.ts

@@ -29,6 +29,7 @@ vi.mock('@google/gemini-cli-core', async (importOriginal) => {
     await importOriginal<typeof import('@google/gemini-cli-core')>();
   return {
     ...actual,
+    PRIORITY_YOLO_ALLOW_ALL: 998,
     Config: vi.fn().mockImplementation((params) => {
       const mockConfig = {
         ...params,

packages/a2a-server/src/config/config.ts

@@ -87,6 +87,7 @@ export async function loadConfig(
         approvalMode === ApprovalMode.YOLO
           ? [
               {
+                toolName: '*',
                 decision: PolicyDecision.ALLOW,
                 priority: PRIORITY_YOLO_ALLOW_ALL,
                 modes: [ApprovalMode.YOLO],

packages/a2a-server/src/utils/testing_utils.ts

@@ -97,6 +97,7 @@ export function createMockConfig(
     getMcpClientManager: vi.fn().mockReturnValue({
       getMcpServers: vi.fn().mockReturnValue({}),
     }),
+    getTelemetryLogPromptsEnabled: vi.fn().mockReturnValue(false),
     getGitService: vi.fn(),
     validatePathAccess: vi.fn().mockReturnValue(undefined),
     getShellExecutionConfig: vi.fn().mockReturnValue({

packages/cli/GEMINI.md

@@ -7,7 +7,10 @@
 - **Shortcuts**: only define keyboard shortcuts in
   `packages/cli/src/ui/key/keyBindings.ts`
 - Do not implement any logic performing custom string measurement or string
-  truncation. Use Ink layout instead leveraging ResizeObserver as needed.
+  truncation. Use Ink layout instead leveraging ResizeObserver as needed. When
+  using `ResizeObserver`, prefer the `useCallback` ref pattern (as seen in
+  `MaxSizedBox.tsx`) to ensure size measurements are captured as soon as the
+  element is available, avoiding potential rendering timing issues.
 - Avoid prop drilling when at all possible.
 
 ## Testing

packages/cli/index.ts

@@ -6,12 +6,19 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import { main } from './src/gemini.js';
-import { FatalError, writeToStderr } from '@google/gemini-cli-core';
-import { runExitCleanup } from './src/utils/cleanup.js';
+// --- Fast Path for Version ---
+// We check for version flags at the very top to avoid loading any heavy dependencies.
+// process.env.CLI_VERSION is defined during the build process by esbuild.
+if (process.argv.includes('--version') || process.argv.includes('-v')) {
+  console.log(process.env['CLI_VERSION'] || 'unknown');
+  process.exit(0);
+}
 
 // --- Global Entry Point ---
 
+let writeToStderrFn: (message: string) => void = (msg) =>
+  process.stderr.write(msg);
+
 // Suppress known race condition error in node-pty on Windows
 // Tracking bug: https://github.com/microsoft/node-pty/issues/827
 process.on('uncaughtException', (error) => {
@@ -28,13 +35,22 @@ process.on('uncaughtException', (error) => {
   // For other errors, we rely on the default behavior, but since we attached a listener,
   // we must manually replicate it.
   if (error instanceof Error) {
-    writeToStderr(error.stack + '\n');
+    writeToStderrFn(error.stack + '\n');
   } else {
-    writeToStderr(String(error) + '\n');
+    writeToStderrFn(String(error) + '\n');
   }
   process.exit(1);
 });
 
+const [{ main }, { FatalError, writeToStderr }, { runExitCleanup }] =
+  await Promise.all([
+    import('./src/gemini.js'),
+    import('@google/gemini-cli-core'),
+    import('./src/utils/cleanup.js'),
+  ]);
+
+writeToStderrFn = writeToStderr;
+
 main().catch(async (error) => {
   // Set a timeout to force exit if cleanup hangs
   const cleanupTimeout = setTimeout(() => {

packages/cli/package.json

@@ -30,7 +30,7 @@
     "sandboxImageUri": "us-docker.pkg.dev/gemini-code-dev/gemini-cli/sandbox:0.36.0-nightly.20260317.2f90b4653"
   },
   "dependencies": {
-    "@agentclientprotocol/sdk": "^0.12.0",
+    "@agentclientprotocol/sdk": "^0.16.1",
     "@google/gemini-cli-core": "file:../core",
     "@google/genai": "1.30.0",
     "@iarna/toml": "^2.2.5",

packages/cli/src/acp/acpClient.test.ts

@@ -551,7 +551,7 @@ describe('GeminiAgent', () => {
     });
 
     expect(session.prompt).toHaveBeenCalled();
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
   });
 
   it('should delegate setMode to session', async () => {
@@ -750,7 +750,7 @@ describe('Session', () => {
         content: { type: 'text', text: 'Hello' },
       },
     });
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
   });
 
   it('should handle /memory command', async () => {
@@ -767,7 +767,7 @@ describe('Session', () => {
       prompt: [{ type: 'text', text: '/memory view' }],
     });
 
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
     expect(handleCommandSpy).toHaveBeenCalledWith(
       '/memory view',
       expect.any(Object),
@@ -789,7 +789,7 @@ describe('Session', () => {
       prompt: [{ type: 'text', text: '/extensions list' }],
     });
 
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
     expect(handleCommandSpy).toHaveBeenCalledWith(
       '/extensions list',
       expect.any(Object),
@@ -811,7 +811,7 @@ describe('Session', () => {
       prompt: [{ type: 'text', text: '/extensions explore' }],
     });
 
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
     expect(handleCommandSpy).toHaveBeenCalledWith(
       '/extensions explore',
       expect.any(Object),
@@ -833,7 +833,7 @@ describe('Session', () => {
       prompt: [{ type: 'text', text: '/restore' }],
     });
 
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
     expect(handleCommandSpy).toHaveBeenCalledWith(
       '/restore',
       expect.any(Object),
@@ -855,7 +855,7 @@ describe('Session', () => {
       prompt: [{ type: 'text', text: '/init' }],
     });
 
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
     expect(handleCommandSpy).toHaveBeenCalledWith('/init', expect.any(Object));
     expect(mockChat.sendMessageStream).not.toHaveBeenCalled();
   });
@@ -909,7 +909,7 @@ describe('Session', () => {
         }),
       }),
     );
-    expect(result).toEqual({ stopReason: 'end_turn' });
+    expect(result).toMatchObject({ stopReason: 'end_turn' });
   });
 
   it('should handle tool call permission request', async () => {
@@ -1080,6 +1080,70 @@ describe('Session', () => {
     );
   });
 
+  it('should split getDisplayTitle and getExplanation for title and content in permission request', async () => {
+    const confirmationDetails = {
+      type: 'info',
+      onConfirm: vi.fn(),
+    };
+    mockTool.build.mockReturnValue({
+      getDescription: () => 'Original Description',
+      getDisplayTitle: () => 'Display Title Only',
+      getExplanation: () => 'A detailed explanation text',
+      toolLocations: () => [],
+      shouldConfirmExecute: vi.fn().mockResolvedValue(confirmationDetails),
+      execute: vi.fn().mockResolvedValue({ llmContent: 'Tool Result' }),
+    });
+
+    mockConnection.requestPermission.mockResolvedValue({
+      outcome: {
+        outcome: 'selected',
+        optionId: ToolConfirmationOutcome.ProceedOnce,
+      },
+    });
+
+    const stream1 = createMockStream([
+      {
+        type: StreamEventType.CHUNK,
+        value: {
+          functionCalls: [{ name: 'test_tool', args: {} }],
+        },
+      },
+    ]);
+    const stream2 = createMockStream([
+      {
+        type: StreamEventType.CHUNK,
+        value: { candidates: [] },
+      },
+    ]);
+
+    mockChat.sendMessageStream
+      .mockResolvedValueOnce(stream1)
+      .mockResolvedValueOnce(stream2);
+
+    await session.prompt({
+      sessionId: 'session-1',
+      prompt: [{ type: 'text', text: 'Call tool' }],
+    });
+
+    expect(mockConnection.requestPermission).toHaveBeenCalledWith(
+      expect.objectContaining({
+        toolCall: expect.objectContaining({
+          title: 'Display Title Only',
+          content: [],
+        }),
+      }),
+    );
+
+    expect(mockConnection.sessionUpdate).toHaveBeenCalledWith(
+      expect.objectContaining({
+        update: expect.objectContaining({
+          sessionUpdate: 'agent_thought_chunk',
+          content: { type: 'text', text: 'A detailed explanation text' },
+        }),
+      }),
+    );
+  });
+
   it('should use filePath for ACP diff content in tool result', async () => {
     mockTool.build.mockReturnValue({
       getDescription: () => 'Test Tool',

packages/cli/src/acp/acpClient.ts

@@ -98,6 +98,12 @@ export async function runAcpClient(
 }
 
 export class GeminiAgent {
+  private static callIdCounter = 0;
+
+  static generateCallId(name: string): string {
+    return `${name}-${Date.now()}-${++GeminiAgent.callIdCounter}`;
+  }
+
   private sessions: Map<string, Session> = new Map();
   private clientCapabilities: acp.ClientCapabilities | undefined;
   private apiKey: string | undefined;
@@ -294,6 +300,7 @@ export class GeminiAgent {
         sessionId,
         this.clientCapabilities.fs,
         config.getFileSystemService(),
+        cwd,
       );
       config.setFileSystemService(acpFileSystemService);
     }
@@ -351,16 +358,6 @@ export class GeminiAgent {
     const { sessionData, sessionPath } =
       await sessionSelector.resolveSession(sessionId);
 
-    if (this.clientCapabilities?.fs) {
-      const acpFileSystemService = new AcpFileSystemService(
-        this.connection,
-        sessionId,
-        this.clientCapabilities.fs,
-        config.getFileSystemService(),
-      );
-      config.setFileSystemService(acpFileSystemService);
-    }
-
     const clientHistory = convertSessionToClientHistory(sessionData.messages);
 
     const geminiClient = config.getGeminiClient();
@@ -434,7 +431,19 @@ export class GeminiAgent {
       throw acp.RequestError.authRequired();
     }
 
-    // 3. Now that we are authenticated, it is safe to initialize the config
+    // 3. Set the ACP FileSystemService (if supported) before config initialization
+    if (this.clientCapabilities?.fs) {
+      const acpFileSystemService = new AcpFileSystemService(
+        this.connection,
+        sessionId,
+        this.clientCapabilities.fs,
+        config.getFileSystemService(),
+        cwd,
+      );
+      config.setFileSystemService(acpFileSystemService);
+    }
+
+    // 4. Now that we are authenticated, it is safe to initialize the config
     // which starts the MCP servers and other heavy resources.
     await config.initialize();
     startupProfiler.flush(config);
@@ -699,10 +708,22 @@ export class Session {
       // It uses `parts` argument but effectively ignores it in current implementation
       const handled = await this.handleCommand(commandText, parts);
       if (handled) {
-        return { stopReason: 'end_turn' };
+        return {
+          stopReason: 'end_turn',
+          _meta: {
+            quota: {
+              token_count: { input_tokens: 0, output_tokens: 0 },
+              model_usage: [],
+            },
+          },
+        };
       }
     }
 
+    let totalInputTokens = 0;
+    let totalOutputTokens = 0;
+    const modelUsageMap = new Map<string, { input: number; output: number }>();
+
     let nextMessage: Content | null = { role: 'user', parts };
 
     while (nextMessage !== null) {
@@ -727,11 +748,25 @@ export class Session {
         );
         nextMessage = null;
 
+        let turnInputTokens = 0;
+        let turnOutputTokens = 0;
+        let turnModelId = model;
+
         for await (const resp of responseStream) {
           if (pendingSend.signal.aborted) {
             return { stopReason: CoreToolCallStatus.Cancelled };
           }
 
+          if (resp.type === StreamEventType.CHUNK && resp.value.usageMetadata) {
+            turnInputTokens =
+              resp.value.usageMetadata.promptTokenCount ?? turnInputTokens;
+            turnOutputTokens =
+              resp.value.usageMetadata.candidatesTokenCount ?? turnOutputTokens;
+            if (resp.value.modelVersion) {
+              turnModelId = resp.value.modelVersion;
+            }
+          }
+
           if (
             resp.type === StreamEventType.CHUNK &&
             resp.value.candidates &&
@@ -763,6 +798,19 @@ export class Session {
           }
         }
 
+        totalInputTokens += turnInputTokens;
+        totalOutputTokens += turnOutputTokens;
+
+        if (turnInputTokens > 0 || turnOutputTokens > 0) {
+          const existing = modelUsageMap.get(turnModelId) ?? {
+            input: 0,
+            output: 0,
+          };
+          existing.input += turnInputTokens;
+          existing.output += turnOutputTokens;
+          modelUsageMap.set(turnModelId, existing);
+        }
+
         if (pendingSend.signal.aborted) {
           return { stopReason: CoreToolCallStatus.Cancelled };
         }
@@ -799,7 +847,28 @@ export class Session {
       }
     }
 
-    return { stopReason: 'end_turn' };
+    const modelUsageArray = Array.from(modelUsageMap.entries()).map(
+      ([modelName, counts]) => ({
+        model: modelName,
+        token_count: {
+          input_tokens: counts.input,
+          output_tokens: counts.output,
+        },
+      }),
+    );
+
+    return {
+      stopReason: 'end_turn',
+      _meta: {
+        quota: {
+          token_count: {
+            input_tokens: totalInputTokens,
+            output_tokens: totalOutputTokens,
+          },
+          model_usage: modelUsageArray,
+        },
+      },
+    };
   }
 
   private async handleCommand(
@@ -837,7 +906,7 @@ export class Session {
     promptId: string,
     fc: FunctionCall,
   ): Promise<Part[]> {
-    const callId = fc.id ?? `${fc.name}-${Date.now()}`;
+    const callId = fc.id ?? GeminiAgent.generateCallId(fc.name || 'unknown');
     const args = fc.args ?? {};
 
     const startTime = Date.now();
@@ -887,6 +956,23 @@ export class Session {
     try {
       const invocation = tool.build(args);
 
+      const displayTitle =
+        typeof invocation.getDisplayTitle === 'function'
+          ? invocation.getDisplayTitle()
+          : invocation.getDescription();
+
+      const explanation =
+        typeof invocation.getExplanation === 'function'
+          ? invocation.getExplanation()
+          : '';
+
+      if (explanation) {
+        await this.sendUpdate({
+          sessionUpdate: 'agent_thought_chunk',
+          content: { type: 'text', text: explanation },
+        });
+      }
+
       const confirmationDetails =
         await invocation.shouldConfirmExecute(abortSignal);
 
@@ -918,7 +1004,7 @@ export class Session {
           toolCall: {
             toolCallId: callId,
             status: 'pending',
-            title: invocation.getDescription(),
+            title: displayTitle,
             content,
             locations: invocation.toolLocations(),
             kind: toAcpToolKind(tool.kind),
@@ -954,12 +1040,14 @@ export class Session {
           }
         }
       } else {
+        const content: acp.ToolCallContent[] = [];
+
         await this.sendUpdate({
           sessionUpdate: 'tool_call',
           toolCallId: callId,
           status: 'in_progress',
-          title: invocation.getDescription(),
-          content: [],
+          title: displayTitle,
+          content,
           locations: invocation.toolLocations(),
           kind: toAcpToolKind(tool.kind),
         });
@@ -968,12 +1056,14 @@ export class Session {
       const toolResult: ToolResult = await invocation.execute(abortSignal);
       const content = toToolCallContent(toolResult);
 
+      const updateContent: acp.ToolCallContent[] = content ? [content] : [];
+
       await this.sendUpdate({
         sessionUpdate: 'tool_call_update',
         toolCallId: callId,
         status: 'completed',
-        title: invocation.getDescription(),
-        content: content ? [content] : [],
+        title: displayTitle,
+        content: updateContent,
         locations: invocation.toolLocations(),
         kind: toAcpToolKind(tool.kind),
       });
@@ -1310,7 +1400,7 @@ export class Session {
         include: pathSpecsToRead,
       };
 
-      const callId = `${readManyFilesTool.name}-${Date.now()}`;
+      const callId = GeminiAgent.generateCallId(readManyFilesTool.name);
 
       try {
         const invocation = readManyFilesTool.build(toolArgs);
@@ -1538,6 +1628,7 @@ function toPermissionOptions(
     case 'info':
     case 'ask_user':
     case 'exit_plan_mode':
+    case 'sandbox_expansion':
       break;
     default: {
       const unreachable: never = confirmation;

packages/cli/src/acp/fileSystemService.test.ts

@@ -4,10 +4,25 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import { describe, it, expect, vi, beforeEach, type Mocked } from 'vitest';
+import {
+  describe,
+  it,
+  expect,
+  vi,
+  beforeEach,
+  afterEach,
+  type Mocked,
+} from 'vitest';
 import { AcpFileSystemService } from './fileSystemService.js';
 import type { AgentSideConnection } from '@agentclientprotocol/sdk';
 import type { FileSystemService } from '@google/gemini-cli-core';
+import os from 'node:os';
+
+vi.mock('node:os', () => ({
+  default: {
+    homedir: vi.fn(),
+  },
+}));
 
 describe('AcpFileSystemService', () => {
   let mockConnection: Mocked<AgentSideConnection>;
@@ -25,13 +40,19 @@ describe('AcpFileSystemService', () => {
       readTextFile: vi.fn(),
       writeTextFile: vi.fn(),
     };
+    vi.mocked(os.homedir).mockReturnValue('/home/user');
+  });
+
+  afterEach(() => {
+    vi.restoreAllMocks();
   });
 
   describe('readTextFile', () => {
     it.each([
       {
         capability: true,
-        desc: 'connection if capability exists',
+        path: '/path/to/file',
+        desc: 'connection if capability exists and file is inside root',
         setup: () => {
           mockConnection.readTextFile.mockResolvedValue({ content: 'content' });
         },
@@ -45,6 +66,7 @@ describe('AcpFileSystemService', () => {
       },
       {
         capability: false,
+        path: '/path/to/file',
         desc: 'fallback if capability missing',
         setup: () => {
           mockFallback.readTextFile.mockResolvedValue('content');
@@ -56,19 +78,72 @@ describe('AcpFileSystemService', () => {
           expect(mockConnection.readTextFile).not.toHaveBeenCalled();
         },
       },
-    ])('should use $desc', async ({ capability, setup, verify }) => {
+      {
+        capability: true,
+        path: '/outside/file',
+        desc: 'fallback if capability exists but file is outside root',
+        setup: () => {
+          mockFallback.readTextFile.mockResolvedValue('content');
+        },
+        verify: () => {
+          expect(mockFallback.readTextFile).toHaveBeenCalledWith(
+            '/outside/file',
+          );
+          expect(mockConnection.readTextFile).not.toHaveBeenCalled();
+        },
+      },
+      {
+        capability: true,
+        path: '/home/user/.gemini/tmp/file.md',
+        root: '/home/user',
+        desc: 'fallback if file is inside global gemini dir, even if root overlaps',
+        setup: () => {
+          mockFallback.readTextFile.mockResolvedValue('content');
+        },
+        verify: () => {
+          expect(mockFallback.readTextFile).toHaveBeenCalledWith(
+            '/home/user/.gemini/tmp/file.md',
+          );
+          expect(mockConnection.readTextFile).not.toHaveBeenCalled();
+        },
+      },
+    ])(
+      'should use $desc',
+      async ({ capability, path, root, setup, verify }) => {
+        service = new AcpFileSystemService(
+          mockConnection,
+          'session-1',
+          { readTextFile: capability, writeTextFile: true },
+          mockFallback,
+          root || '/path/to',
+        );
+        setup();
+
+        const result = await service.readTextFile(path);
+
+        expect(result).toBe('content');
+        verify();
+      },
+    );
+
+    it('should throw normalized ENOENT error when readTextFile encounters "Resource not found"', async () => {
       service = new AcpFileSystemService(
         mockConnection,
         'session-1',
-        { readTextFile: capability, writeTextFile: true },
+        { readTextFile: true, writeTextFile: true },
         mockFallback,
+        '/path/to',
+      );
+      mockConnection.readTextFile.mockRejectedValue(
+        new Error('Resource not found for document'),
       );
-      setup();
 
-      const result = await service.readTextFile('/path/to/file');
-
-      expect(result).toBe('content');
-      verify();
+      await expect(
+        service.readTextFile('/path/to/missing'),
+      ).rejects.toMatchObject({
+        code: 'ENOENT',
+        message: 'Resource not found for document',
+      });
     });
   });
 
@@ -76,7 +151,8 @@ describe('AcpFileSystemService', () => {
     it.each([
       {
         capability: true,
-        desc: 'connection if capability exists',
+        path: '/path/to/file',
+        desc: 'connection if capability exists and file is inside root',
         verify: () => {
           expect(mockConnection.writeTextFile).toHaveBeenCalledWith({
             path: '/path/to/file',
@@ -88,6 +164,7 @@ describe('AcpFileSystemService', () => {
       },
       {
         capability: false,
+        path: '/path/to/file',
         desc: 'fallback if capability missing',
         verify: () => {
           expect(mockFallback.writeTextFile).toHaveBeenCalledWith(
@@ -97,17 +174,63 @@ describe('AcpFileSystemService', () => {
           expect(mockConnection.writeTextFile).not.toHaveBeenCalled();
         },
       },
-    ])('should use $desc', async ({ capability, verify }) => {
+      {
+        capability: true,
+        path: '/outside/file',
+        desc: 'fallback if capability exists but file is outside root',
+        verify: () => {
+          expect(mockFallback.writeTextFile).toHaveBeenCalledWith(
+            '/outside/file',
+            'content',
+          );
+          expect(mockConnection.writeTextFile).not.toHaveBeenCalled();
+        },
+      },
+      {
+        capability: true,
+        path: '/home/user/.gemini/tmp/file.md',
+        root: '/home/user',
+        desc: 'fallback if file is inside global gemini dir, even if root overlaps',
+        verify: () => {
+          expect(mockFallback.writeTextFile).toHaveBeenCalledWith(
+            '/home/user/.gemini/tmp/file.md',
+            'content',
+          );
+          expect(mockConnection.writeTextFile).not.toHaveBeenCalled();
+        },
+      },
+    ])('should use $desc', async ({ capability, path, root, verify }) => {
       service = new AcpFileSystemService(
         mockConnection,
         'session-1',
         { writeTextFile: capability, readTextFile: true },
         mockFallback,
+        root || '/path/to',
       );
 
-      await service.writeTextFile('/path/to/file', 'content');
+      await service.writeTextFile(path, 'content');
 
       verify();
     });
+
+    it('should throw normalized ENOENT error when writeTextFile encounters "Resource not found"', async () => {
+      service = new AcpFileSystemService(
+        mockConnection,
+        'session-1',
+        { readTextFile: true, writeTextFile: true },
+        mockFallback,
+        '/path/to',
+      );
+      mockConnection.writeTextFile.mockRejectedValue(
+        new Error('Resource not found for directory'),
+      );
+
+      await expect(
+        service.writeTextFile('/path/to/missing', 'content'),
+      ).rejects.toMatchObject({
+        code: 'ENOENT',
+        message: 'Resource not found for directory',
+      });
+    });
   });
 });

packages/cli/src/acp/fileSystemService.ts

@@ -4,44 +4,82 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import type { FileSystemService } from '@google/gemini-cli-core';
+import { isWithinRoot, type FileSystemService } from '@google/gemini-cli-core';
 import type * as acp from '@agentclientprotocol/sdk';
+import os from 'node:os';
+import path from 'node:path';
 
 /**
  * ACP client-based implementation of FileSystemService
  */
 export class AcpFileSystemService implements FileSystemService {
+  private readonly geminiDir = path.join(os.homedir(), '.gemini');
+
   constructor(
     private readonly connection: acp.AgentSideConnection,
     private readonly sessionId: string,
-    private readonly capabilities: acp.FileSystemCapability,
+    private readonly capabilities: acp.FileSystemCapabilities,
     private readonly fallback: FileSystemService,
+    private readonly root: string,
   ) {}
 
+  private shouldUseFallback(filePath: string): boolean {
+    // Files inside the global CLI directory must always use the native file system,
+    // even if the user runs the CLI directly from their home directory (which
+    // would make the IDE's project root overlap with the global directory).
+    return (
+      !isWithinRoot(filePath, this.root) ||
+      isWithinRoot(filePath, this.geminiDir)
+    );
+  }
+
+  private normalizeFileSystemError(err: unknown): never {
+    const errorMessage = err instanceof Error ? err.message : String(err);
+    if (
+      errorMessage.includes('Resource not found') ||
+      errorMessage.includes('ENOENT') ||
+      errorMessage.includes('does not exist') ||
+      errorMessage.includes('No such file')
+    ) {
+      const newErr = new Error(errorMessage) as NodeJS.ErrnoException;
+      newErr.code = 'ENOENT';
+      throw newErr;
+    }
+    throw err;
+  }
+
   async readTextFile(filePath: string): Promise<string> {
-    if (!this.capabilities.readTextFile) {
+    if (!this.capabilities.readTextFile || this.shouldUseFallback(filePath)) {
       return this.fallback.readTextFile(filePath);
     }
 
-    // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
-    const response = await this.connection.readTextFile({
-      path: filePath,
-      sessionId: this.sessionId,
-    });
+    try {
+      // eslint-disable-next-line @typescript-eslint/no-unsafe-assignment
+      const response = await this.connection.readTextFile({
+        path: filePath,
+        sessionId: this.sessionId,
+      });
 
-    // eslint-disable-next-line @typescript-eslint/no-unsafe-return
-    return response.content;
+      // eslint-disable-next-line @typescript-eslint/no-unsafe-return
+      return response.content;
+    } catch (err: unknown) {
+      this.normalizeFileSystemError(err);
+    }
   }
 
   async writeTextFile(filePath: string, content: string): Promise<void> {
-    if (!this.capabilities.writeTextFile) {
+    if (!this.capabilities.writeTextFile || this.shouldUseFallback(filePath)) {
       return this.fallback.writeTextFile(filePath, content);
     }
 
-    await this.connection.writeTextFile({
-      path: filePath,
-      content,
-      sessionId: this.sessionId,
-    });
+    try {
+      await this.connection.writeTextFile({
+        path: filePath,
+        content,
+        sessionId: this.sessionId,
+      });
+    } catch (err: unknown) {
+      this.normalizeFileSystemError(err);
+    }
   }
 }

packages/cli/src/commands/extensions/examples/policies/policies/policies.toml

@@ -16,7 +16,7 @@ toolName = "grep_search"
 argsPattern = "(\.env|id_rsa|passwd)"
 decision = "deny"
 priority = 200
-deny_message = "Access to sensitive credentials or system files is restricted by the policy-example extension."
+denyMessage = "Access to sensitive credentials or system files is restricted by the policy-example extension."
 
 # Safety Checker: Apply path validation to all write operations.
 [[safety_checker]]