mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-07 16:11:46 +02:00
DottaandPaperclip
c481be44e3
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Task watchdogs review issue subtrees when no run or queued wake keeps work live > - Pending human interactions and approvals are valid stopped states that still need one watchdog review > - The existing fingerprint included volatile activity timestamps, so unchanged stopped trees could wake repeatedly after comments, documents, work products, or sibling completions > - This pull request fingerprints only review-material leaf and wait state, persists the reviewed snapshot, and suppresses shrink-only repeats > - The benefit is one review per materially new stopped state without weakening liveness classification or hiding human waits ## Linked Issues or Issue Description ### What happened? Task-watchdog stop fingerprints changed for metadata-only activity and completed siblings, producing duplicate wakes after an unchanged stop had already been reviewed. ### Expected behavior Pending interactions and approvals remain classified as stopped, but a reviewed stopped state only wakes again when waits, non-terminal leaves, status, assignment, or blockers gain material changes. ### Steps to reproduce 1. Review a stopped watched subtree with a pending human wait or multiple non-terminal stopped leaves. 2. Add only comment/document/work-product activity, or complete one stopped sibling without changing the wait set. 3. Observe a duplicate wake from the timestamp-heavy fingerprint. Related public work: Refs #9452 for overlapping task-watchdog service edits and #10043 for related no-op fingerprint suppression. ## What Changed - Added fingerprint v2 over non-terminal material leaves plus subtree-wide pending wait ids, excluding volatile timestamps while retaining them in wake context. - Added nullable observed/reviewed JSONB stop snapshots and shrink-only reviewed-state suppression with legacy exact-fingerprint fallback. - Added pending interaction kinds and approval ids to watchdog wake context, review comments, and comment metadata. - Added classifier and scheduler coverage for waiting-leaf liveness, metadata stability, sibling shrink suppression, material changes, snapshot promotion, legacy rows, and unchanged idempotency keys. ## Verification - `pnpm --filter @paperclipai/server exec vitest run src/__tests__/task-watchdogs-classifier.test.ts src/__tests__/task-watchdogs-scheduler.test.ts` — 2 files, 36 tests passed. - `pnpm --filter @paperclipai/server typecheck` — passed. - `git diff --check origin/master...HEAD` — passed. ## Risks - Fingerprint version 2 intentionally re-fingerprints every currently stopped watched tree once after deployment, causing a one-time wake burst before the new reviewed snapshots are established. - Migration `0191_task_watchdog_stop_snapshots.sql` only adds two nullable JSONB columns with no backfill; legacy rows continue exact-fingerprint behavior until a post-deploy review promotes a snapshot. - PR #9452 edits the same service file; whichever lands second may need a trivial rebase. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - OpenAI Codex, exact model ID `gpt-5.6-sol`, high reasoning mode, with repository tool use and code execution. The runtime did not expose a context-window size. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing>