DOC-7085 Replace assess-comments ledger with repo-memory precedent lookup

[andy-stark-redis]

Summary

• Removes the hand-maintained coverage ledger (.claude/state/assess-comments.coverage.md) from /docs:assess-comments — it was a single shared file every branch's run wrote to, so it repeatedly collided on merges (documented twice in its own worked-examples log), and it duplicated by hand what /reflect + /finalize squash messages already carry.
• Adds step 3a: treats the Redis repo memory bot's PR comment as a live pointer into project history instead. Decodes the comment's hidden redis-repo-memory-data blob (score + ~500-char preview per related item, already fetched at zero extra cost) to rank relevance, then follows sourceUrl to pull the real squash commit message only for items worth citing as precedent during adjudication.
• Folds the ledger's durable, cross-PR bugbot calibration axes (async dispatch order, npm transitive-dep hoisting, list-vs-item endpoint shape, trusting a stale PR description) into a short static reference block in step 7, since those aren't tied to any one PR and repo memory's per-PR retrieval won't automatically resurface them as a general rule.

Test plan

• Dry-ran the new step 3a logic against PR DOC-7027 Document StackExchange.Redis smart client handoffs (3.3.0) #4047's real repo-memory comment — confirmed the visible bullets only echo the title, decoded the hidden data blob to verify score/bodySummary are present and useful, and confirmed the preview is a genuine (if truncated) slice of a real /reflect+/finalize squash message.
• Run /docs:assess-comments against this PR once Bugbot and repo memory have both commented, to see the full step 1–9 flow end to end.

---

Note

Low Risk
Documentation-only changes to a Claude command skill; no runtime code, auth, or data paths affected.

Overview
Replaces the hand-maintained assess-comments coverage ledger with Redis repo memory as the source of cross-PR precedent, and makes /docs:assess-comments fully report-only with no disk writes.

Removed: .claude/state/assess-comments.coverage.md and all steps that read/updated it (former step 1 ledger load and step 11 evidence-gated updates). The history commenter role is dropped from reconciliation.

Added: A repo-memory source role and step 3a, which decodes the bot comment’s hidden redis-repo-memory-data blob (score-ranked previews), then optionally fetches merged PR squash messages for adjudication precedent. Step 7 gains a static known bugbot calibration patterns block (moved from the ledger’s worked examples). Reports now cite repo-memory PR numbers in verdicts and list which precedent PRs were opened in “Story so far.”

Bot calibration from resolved threads is noted in-run only; persistence is delegated to repo memory and /reflect+/finalize squash messages instead of a shared file that collided on merges.

^{Reviewed by Cursor Bugbot for commit 68da99b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

DOC-7085

[github-actions]

🧠 Redis Memory

Found 5 related items from repository history (5 new this commit):

• Commit — Update assess-comments coverage ledger (PR #3714 encounter) 🆕

  Update assess-comments coverage ledger (PR DOC-6887 Add hash field expiration examples to hash data type page #3714 encounter)

• Commit — DOC-6909 Teach the feed pipeline and version archiver the /content/ link form 🆕

  DOC-6909 Teach the feed pipeline and version archiver the /content/ link form

• Commit — DOC-6909 Teach the feed pipeline and version archiver the /content/ link form 🆕

  DOC-6909 Teach the feed pipeline and version archiver the /content/ link form

• Commit — DOC-6909 Teach the feed pipeline and version archiver the /content/ link form 🆕

  DOC-6909 Teach the feed pipeline and version archiver the /content/ link form

• Commit — DOC-6909 Teach the feed pipeline and version archiver the /content/ link form 🆕

  DOC-6909 Teach the feed pipeline and version archiver the /content/ link form

Memory updated at 68da99b

[andy-stark-redis]

Thanks once again, @dwdougherty ! Have a great weekend :-)