diff --git a/.github/workflows/gemini-cli-bot-brain.yml b/.github/workflows/gemini-cli-bot-brain.yml
index dab5cd1133..cbff0f2fa5 100644
--- a/.github/workflows/gemini-cli-bot-brain.yml
+++ b/.github/workflows/gemini-cli-bot-brain.yml
@@ -44,6 +44,11 @@ jobs:
         env:
           GITHUB_TOKEN: '${{ secrets.GITHUB_TOKEN }}'
         run: |
+          if [ "${{ github.event.inputs.clear_memory }}" = "true" ]; then
+            echo "Memory clear requested. Skipping previous state download."
+            exit 0
+          fi
+
           # Find the last successful run of this workflow
           LAST_RUN_ID=$(gh run list --workflow "${{ github.workflow }}" --status success --limit 1 --json databaseId --jq '.[0].databaseId')
 
@@ -73,6 +78,7 @@ jobs:
           GEMINI_API_KEY: '${{ secrets.GEMINI_API_KEY }}'
           GITHUB_TOKEN: '${{ secrets.GITHUB_TOKEN }}'
           GEMINI_MODEL: 'gemini-3-flash-preview'
+          ENABLE_PRS: '${{ github.event.inputs.enable_prs || 'false' }}'
         run: 'node bundle/gemini.js --policy tools/gemini-cli-bot/ci-policy.toml tools/gemini-cli-bot/brain/metrics.md'
 
       - name: 'Stash Brain Outputs'
@@ -112,3 +118,5 @@ jobs:
           name: 'brain-scripts'
           path: 'temp_outputs/reflexes/scripts/'
           retention-days: 90
+_outputs/reflexes/scripts/'
+          retention-days: 90
diff --git a/tools/gemini-cli-bot/brain/metrics.md b/tools/gemini-cli-bot/brain/metrics.md
index df3649bc0b..cb9e3b9e50 100644
--- a/tools/gemini-cli-bot/brain/metrics.md
+++ b/tools/gemini-cli-bot/brain/metrics.md
@@ -14,79 +14,45 @@ repository health.
   `tools/gemini-cli-bot/history/metrics-before-prev.csv` and the current run's
   metrics.
 - Findings and state are recorded in `tools/gemini-cli-bot/lessons-learned.md`.
+- **Preservation Status**: Check the `ENABLE_PRS` environment variable. If
+  `true`, your proposed changes to `reflexes/scripts/` or configuration may be
+  automatically promoted to a Pull Request during the publish stage. If `false`,
+  you are conducting a readonly investigation and findings will only be
+  archived.
 
 ## Repo Policy Priorities
 
-When analyzing data and proposing solutions, prioritize the following in order:
-
-1.  **Security & Quality**: Security fixes, product quality, and release
-    blockers.
-2.  **Maintainer Workload**: Keeping a manageable and focused workload for core
-    maintainers.
-3.  **Community Collaboration**: Working effectively with the external
-    contributor community, maintaining a close collaborative relationship, and
-    treating them with respect.
+... (rest of priorities) ...
 
 ## Instructions
 
 ### 1. Read & Identify Trends (Time-Series Analysis)
 
-- Load and analyze `tools/gemini-cli-bot/history/metrics-timeseries.csv`.
-- Identify significant anomalies or deteriorating trends over time (e.g.,
-  `latency_pr_overall_hours` steadily increasing, `open_issues` growing faster
-  than closure rates, spikes in `review_distribution_variance`).
+... (rest of step 1) ...
 
 ### 2. Hypothesis Testing & Deep Dive
 
-For each metric not meeting goals or showing a negative trend:
-
-- **Develop Competing Hypotheses**: Brainstorm multiple potential root causes
-  (e.g., "PR Latency is high because CI is flaky" vs. "PR Latency is high
-  because reviewers are unresponsive").
-- **Gather Evidence**: Use your tools (e.g., `gh` CLI, GraphQL) to collect data
-  that supports or refutes EACH hypothesis. You may write temporary local
-  scripts to slice the data (e.g., checking issue labels, ages, or assignees).
-- **Select Root Cause**: Identify the hypothesis most strongly supported by the
-  data.
-- **Prioritize Impact**: Always prioritize solving for verified hypotheses that
-  have the largest impact (e.g., if 30 out of 500 PRs have merge conflicts,
-  fixing merge conflicts is lower priority than addressing a bottleneck
-  affecting 300 PRs).
+... (rest of step 2) ...
 
 ### 3. Maintainer Workload Assessment
 
-Before blaming or proposing reflexes that rely on maintainer action (e.g., more
-triage, more reviews):
-
-- **Quantify Capacity**: Assess the volume of open, unactioned work (untriaged
-  issues, review requests) against the number of active maintainers.
-- If the ratio indicates overload, **do not propose solutions that simply
-  generate more pings**. Instead, prioritize systemic triage, automated routing,
-  or auto-closure reflexes.
+... (rest of step 3) ...
 
 ### 4. Actor-Aware Bottleneck Identification
 
-Before proposing an intervention, accurately identify the blocker:
-
-- **Waiting on Author**: Needs a polite nudge or closure grace period.
-- **Waiting on Maintainer**: Needs routing, aggregated reports, or escalation
-  (do not nudge the author).
-- **Waiting on System (CI/Infra)**: Needs tooling fixes or reporting.
+... (rest of step 4) ...
 
 ### 5. Policy Critique & Evaluation
 
-- **Review Existing Policies**: Examine the existing automation in
-  `.github/workflows/` and scripts in `tools/gemini-cli-bot/reflexes/scripts/`.
-- **Analyze Effectiveness**: Based on your metrics analysis, determine if
-  current policies are achieving their goals (e.g., Is triage reducing latency?
-  Are stale issues closed as expected?).
-- **Identify Gaps**: Where is the automation failing? Are there manual tasks
-  that should be automated?
+... (rest of step 5) ...
 
 ### 6. Record Findings & Propose Actions
 
 - Document your formulated hypotheses, the evidence gathered, and your final
   conclusions in `tools/gemini-cli-bot/lessons-learned.md`.
+- **Memory Preservation**: When updating `lessons-learned.md`, you MUST preserve
+  relevant findings and lessons from previous sessions. Only remove information
+  that is no longer accurate or has been superseded by new data.
 - Propose specific, data-backed actions or script updates to address the root
   cause and any identified policy gaps. Ensure proposed actions align with the
   Repo Policy Priorities and include concepts like graceful closures and