feat(capture): passive answer-memory for sessions

[plind-junior]

makes the zero-friction memory loop run end to end: ask a question in one
session and the answer is saved with no human; a fresh session recalls it from
memory without re-reading the project files.

builds directly on #485 (phase-d receipt-gated auto-approve + vouch ingest).
the read path (recall, context-hook) and the auto-approve path already
existed — the only missing link was a passive session trigger, which this adds.

what changed

• new claude-code Stop hook → vouch capture answer.
• capture.last_exchange(transcript_path) — pure extraction of the turn's
  (question, answer), same host-wrapper filtering as first_user_prompt.
• capture.capture_answer(...) — ingests the answer as a content-addressed
  source, files a receipt-backed claim per quotable span via
  extract.extract_receipt_claims, and self-approves only what
  proposals.approve already allows (review.approver_role: trusted-agent or
  review.auto_approve_on_receipt, whose receipts verify by construction).
  with neither opt-in the claims stay pending — no parallel write path, the
  review gate holds.
• idempotent (same answer bytes skipped) and length-guarded (short
  acknowledgements skipped), so the every-turn Stop hook never duplicates or
  floods the kb.

verification

drove the real two-session flow through the cli, not just unit tests: session 1
vouch capture answer → 4 claims filed, 4 auto-approved, 0 pending;
fresh-process recall + context-hook inject the roadmap answer with 0 tool
calls. probes: re-fire → already-captured, short answer → answer-too-short. 12
new unit tests; full pytest + mypy + ruff green.

known limit

selection is coarse — one claim per sentence, same stance as #485. bounded by a
length floor and max-claims; smarter selection is a follow-up quality knob.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 36fd8fd0-9192-4190-b566-6a4278ff5e20

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/passive-answer-memory

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}