Brainerd Ruminate

Mine older repo-scoped session history for durable memory gaps and route to the correct harness-specific flow.

Detect the harness first

1. If the run exposes Brainerd staged-ruminate tools, you are in the Pi branch. Stay inside the Pi tool flow.
2. If CODEX_THREAD_ID is present, you are in the Codex branch. Use the Codex wrapper.
3. If BRAINERD_CLAUDE_SESSION_ID or BRAINERD_CLAUDE_TRANSCRIPT_PATH is present, you are in the Claude branch. Use the Claude wrapper.
4. If none of those signals match, stop and report that harness detection failed. Do not guess.

Pi branch

1. Read the ambient brain context already injected into the run.
2. Determine whether the run is preview or apply from the hidden Brainerd context.
3. In preview mode, call brainerd_repo_sessions, derive findings, and stage them through brainerd_stage_ruminate.
4. In apply mode, call brainerd_get_staged_ruminate first and apply only that staged proposal through brainerd_apply_changes.
5. End with a visible section that starts exactly with Brainerd summary:.

Codex and Claude branches

1. Read brain/index.md and brain/principles.md.
2. Check for a staged preview with the matching wrapper:

./scripts/brainerd-codex.sh staged-ruminate
./scripts/brainerd-claude.sh staged-ruminate

3. If the user clearly wants to apply or discard that preview, do so through the matching wrapper.
4. Otherwise run the matching repo-sessions command, derive a new preview, and write a small JSON payload to /tmp/brainerd-ruminate.json.
5. Stage the preview only through the matching wrapper:

./scripts/brainerd-codex.sh stage-ruminate --input /tmp/brainerd-ruminate.json
./scripts/brainerd-claude.sh stage-ruminate --input /tmp/brainerd-ruminate.json

Guardrails

• Preview is not permission to write.
• Use only repo-scoped session history.
• Do not dump raw transcript excerpts into the brain.
• If no durable finding exists, say so and stop.
• End with a short summary of what was staged, applied, discarded, or left unchanged.