[codex] fix(comments): make durable effects retryable

[kkroo]

Thinking Path

• Paperclip is the control plane for AI-agent companies, where comments can trigger durable operational work such as references, activity, and wakes.
• Idempotent comments must be safe to replay after a request, worker, or replica fails partway through that work.
• The predecessor implementation in fix(comments): durable effect ledger for idempotent comment pipeline (supersedes #836) #903 adds a durable per-comment effect ledger with transactionally enqueued intents and lease-based claims.
• Its review found that an effect could become terminally failed, while the comment was still stamped processed even though that effect never ran.
• This pull request independently succeeds fix(comments): durable effect ledger for idempotent comment pipeline (supersedes #836) #903 by preserving the ledger design while keeping failed effects replayable through database-backed backoff and reconciliation.
• The benefit is that transient delivery failures cannot permanently lose a required comment side effect.

Linked Issues or Issue Description

Refs #903 (App-authored predecessor; this independent successor does not modify or close it). Related internal tracking: BLO-19116.

Problem: after ten failures, a comment effect was parked as failed, excluded from retries, and treated as settled. The parent comment then received idempotency_processed_at although a sink such as wake delivery had not completed.

Expected behavior: every non-processed effect remains outstanding. Repeated failures use durable backoff, including a longer cooldown after the threshold, and the reconciler can reclaim the effect until it completes.

What Changed

• Added the durable issue_comment_effects ledger migration and schema, including next_attempt_at plus a due-work index.
• Ported the transactional keyed-comment effect pipeline, lease claims, reconciliation, and wake idempotency receipts from fix(comments): durable effect ledger for idempotent comment pipeline (supersedes #836) #903 onto current master.
• Changed failure handling so failed is a backoff state rather than terminal settlement; comments stamp processed only after every effect is processed.
• Added regression coverage for exhausted-effect recovery through the actual reconciler path.

Verification

• pnpm --filter @paperclipai/db check:migrations — passed.
• pnpm --filter @paperclipai/server exec vitest run --no-file-parallelism --maxWorkers=1 --reporter=verbose src/services/issue-comment-effects.test.ts — 15 passed.
• pnpm --filter @paperclipai/server exec vitest run --no-file-parallelism --maxWorkers=1 --reporter=verbose src/__tests__/issue-update-comment-wakeup-routes.test.ts src/services/wake-idempotency.test.ts — 25 passed.
• pnpm --filter @paperclipai/db typecheck and pnpm --filter @paperclipai/server typecheck — passed.
• pnpm --filter @paperclipai/db build and pnpm --filter @paperclipai/server build — passed.

Risks

• The migration adds a new ledger table and indexes; it is additive and has no data backfill.
• Failed effects now remain visible as unfinished until recovery, which is intentional and avoids falsely acknowledging lost side effects.
• Backoff delays recovery during sustained downstream outages, but prevent hot-loop retries while retaining eventual replay.

For core feature work, check 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 GPT-5 Codex — tool-assisted reasoning and code execution; the session does not expose a more specific serving-model version or context-window size.

Checklist

• I have included a thinking path that traces from project context to this change
• I have specified the model used (with version and capability details)
• I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work
• I have searched GitHub for duplicate or related PRs and linked them above
• I have described the underlying bug in this PR and linked the predecessor
• I have run tests locally and they pass
• I have added or updated tests where applicable
• If this change affects the UI, I have included before/after screenshots (not applicable: no UI change)
• I have updated relevant documentation to reflect my changes (not applicable: no user-facing contract or command changed)
• I have considered and documented any risks above
• All Paperclip CI gates are green (awaiting GitHub Actions)
• Greptile is 5/5 with no open P2s, recommendations, or follow-ups (awaiting review)
• I will address all Greptile and reviewer comments before requesting merge

[allyblockcast]

🔗 Paperclip issue: BLO-19116

[allyblockcast]

🔗 Paperclip issue: BLO-19116

[kkroo]

Closing as stale — this draft can no longer be landed as a diff.

Measured against master ac91e966:

• 808 commits behind
• 6 conflicting file(s) (git merge-tree --write-tree)

At that distance the surrounding code has moved out from under the patch, so "resolve the conflicts" would mean rewriting it against a codebase it was never written for — not merging it. Reopening is cheap if the branch is still wanted; the commits are untouched.

Closing the PR does not close the underlying issue. Where this PR referenced a BLO issue, that issue remains the record of intent and the place to re-implement from, against current master.

No judgement here on whether the original problem still exists — I did not verify that per-PR, and it should be re-checked before any re-implementation, since some of these may already have been fixed independently.