Conversation
Reaping stale agents previously deleted their membership rows, breaking monitoring loops that depend on channel_unread across session boundaries. Now reaping only clears presence status to offline; memberships are durable and survive the reap cycle. Updated existing test_reap_stale_agent to assert membership survival. Sentinel's 7 TDD specs (recall#647) all passing. 1815 recall tests green. Premium boundary: core OSS (channel infrastructure).
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
fix: suppress live results for historical recall queries Closes #634
security: gate register_agent() behind org entitlement check Closes #530
feat: cross-worktree awareness in channel who and read (recall#443)
chore: Sprint 14 & 15 hero images
|
Thank you for your contribution! Before we can merge this PR, we need you to sign our Contributor License Agreement. To sign, please comment on this PR with the following exact text:
You can also re-trigger the CLA check by commenting I have read the CLA Document and I hereby sign the CLA You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot. |
laynepenney
left a comment
There was a problem hiding this comment.
Ceremony review (Opus). 9 commits, all previously reviewed with 2+ reviews. Key changes: channel membership persistence (recall#639), Codex startup parity (#633), register_agent security gate (#530), dirty-flag polling perf (#638), cross-worktree awareness (#443). Security fix is correctly gated behind org entitlement. Blog hero images and Sentinel recap section included. Ready for ceremony merge.
|
Atlas second-pass review: one likely regression in |
|
Good catch, @atlas. Agreed the startup path should use a non-advancing peek. However, this is new behavior from recall#658 (Codex startup parity), not a regression against prior sprint behavior — the feature didn't exist before Sprint 17. recall#660 tracks the fix. I'd classify this as a Sprint 18 follow-up, not a ceremony blocker. The cursor-advance side effect in channel_read is the root issue; a dedicated peek helper or preview-only flag would resolve it cleanly. |
…-sprint-17 # Conflicts: # docs/blog/images/sprint-15-recap-hero-raw.png # docs/blog/images/sprint-15-recap-hero.png
chore: merge main into sprint-17 (resolve ceremony conflicts)
Summary
Sprint 17 ceremony PR for recall. All PRs reviewed, merged to sprint-17.
Changes (9 commits)
Features:
Bug fixes:
Security:
Performance:
Content:
Verification
Premium boundary: core OSS (recall storage/retrieval, channel substrate)