Feature: Implement V0 Episodic Context Manager

- Re-wrote the monolithic string-based context manipulation logic into an elegant, immutable Episodic IR pipeline.
- Implemented four non-destructive degradation processors: `HistorySquashingProcessor`, `ToolMaskingProcessor`, `BlobDegradationProcessor`, and `SemanticCompressionProcessor`.
- Added dynamic configuration knobs (`budget` and `strategies`) to precisely tune the retention algorithms.
- Implemented a power-user `incrementalGc` flag for maximum context preservation beneath the ceiling.
- Enforced strict typing across the new pipeline, replacing unsafe casts with robust mapping interfaces.
- Added `powerUserProfile` to support features for those wanting a bit
  more quality at the cost of tokens.

packages/cli/src/config/config.test.ts

@@ -21,7 +21,7 @@ import {
   type MCPServerConfig,
   type GeminiCLIExtension,
   Storage,
-  generalistProfile,
+  GENERALIST_PROFILE,
   type ContextManagementConfig,
 } from '@google/gemini-cli-core';
 import { loadCliConfig, parseArguments, type CliArgs } from './config.js';
@@ -2211,7 +2211,7 @@ describe('loadCliConfig context management', () => {
     });
     const config = await loadCliConfig(settings, 'test-session', argv);
     expect(config.getContextManagementConfig()).toStrictEqual(
-      generalistProfile,
+      GENERALIST_PROFILE,
     );
     expect(config.isContextManagementEnabled()).toBe(true);
   });
@@ -2220,24 +2220,19 @@ describe('loadCliConfig context management', () => {
     process.argv = ['node', 'script.js'];
     const argv = await parseArguments(createTestMergedSettings());
     const contextManagementConfig: Partial<ContextManagementConfig> = {
-      historyWindow: {
+      budget: {
+        incrementalGc: false,
         maxTokens: 100_000,
         retainedTokens: 50_000,
+        protectedEpisodes: 1,
+        protectSystemEpisode: true,
       },
-      messageLimits: {
-        normalMaxTokens: 1000,
-        retainedMaxTokens: 10_000,
-        normalizationHeadRatio: 0.25,
-      },
-      tools: {
-        distillation: {
-          maxOutputTokens: 10_000,
-          summarizationThresholdTokens: 15_000,
-        },
-        outputMasking: {
-          protectionThresholdTokens: 30_000,
-          minPrunableThresholdTokens: 10_000,
-          protectLatestTurn: false,
+      strategies: {
+        historySquashing: { maxTokensPerNode: 12000 },
+        toolMasking: { stringLengthThresholdTokens: 10000 },
+        semanticCompression: {
+          nodeThresholdTokens: 5000,
+          compressionModel: 'chat-compression-2.5-flash-lite',
         },
       },
     };

packages/cli/src/config/config.ts

@@ -46,7 +46,8 @@ import {
   type HookEventName,
   type OutputFormat,
   detectIdeFromEnv,
-  generalistProfile,
+  GENERALIST_PROFILE,
+  POWER_USER_PROFILE,
 } from '@google/gemini-cli-core';
 import {
   type Settings,
@@ -886,12 +887,15 @@ export async function loadCliConfig(
 
   const useGeneralistProfile =
     settings.experimental?.generalistProfile ?? false;
+  const usePowerUserProfile =
+    settings.experimental?.powerUserProfile ?? false;
   const useContextManagement =
     settings.experimental?.contextManagement ?? false;
   const contextManagement = {
-    ...(useGeneralistProfile ? generalistProfile : {}),
+    ...(useGeneralistProfile ? GENERALIST_PROFILE : {}),
+    ...(usePowerUserProfile ? POWER_USER_PROFILE : {}),
     ...(useContextManagement ? settings?.contextManagement : {}),
-    enabled: useContextManagement || useGeneralistProfile,
+    enabled: useContextManagement || useGeneralistProfile || usePowerUserProfile,
   };
 
   return new Config({

packages/cli/src/config/settingsSchema.ts

@@ -2149,6 +2149,15 @@ const SETTINGS_SCHEMA = {
           'Replace the built-in save_memory tool with a memory manager subagent that supports adding, removing, de-duplicating, and organizing memories.',
         showInDialog: true,
       },
+      powerUserProfile: {
+        type: 'boolean',
+        label: 'Use the power user profile for massive contexts.',
+        category: 'Experimental',
+        requiresRestart: true,
+        default: false,
+        description: 'Enables continuous minimal GC near the max tokens limit instead of a blocked backbuffer.',
+        showInDialog: true,
+      },
       generalistProfile: {
         type: 'boolean',
         label: 'Use the generalist profile to manage agent contexts.',