docs(handoff): handoff-search-index CLOSE-OUT — the doc said NOT deployed about a live, armed timer

[ZacxDev]

The doc was stale in the one direction that matters: /resume reads it first, and it said

🔴 NOT deployed. No home-manager switch was run. The unit exists in the repo only.

against timer=enabled, Result=success, 4,651 rows live. It also listed 2 of 6 merged PRs and showed all four ranked steps open when all four are done.

Records the close-out:

• All four ranked steps done. Arm the index ✅ · unreadable-docs delete fix ✅ (fix(handoff-index): a repo this run could not read IN FULL is not a repo it may DELETE #1267) · decide the four-spelling shape ✅ (answered, below) · wire a consumer ✅ (feat(resume): the section index gets its first caller — at the moment a probe is about to be re-run #1295).
• The live-Postgres gap is closed — the GENERATED … STORED tsv and the GIN index had never been executed by a server, only pinned as SQL text. Now both, plus the timer's own run (Result=success, wrote 4651 section row(s)), which is the only thing that tests the unit's environment.
• The four-spelling shape is answered: DELETE authority was granted by a negative predicate — the absence of whichever failure had already been seen — rather than a positive demonstration that the run holds a complete replacement. A blacklist re-opens the hole for every unmet failure, which is why four rounds never converged. Closed at one level; the residual one level up (authority per derivation, DELETE per label) stays open with a tripwire that fails the day it is fixed.
• Both ladders (5 rounds on feat(handoff-search): section-grained full-text index of the handoff corpus (P1) #1209, 3 on fix(handoff-index): a repo this run could not read IN FULL is not a repo it may DELETE #1267) were stopped on the prose-payload criterion, not a clean round — recorded so it doesn't read as convergence.

Status sections replace, findings append, so the earlier diagnosis text survives verbatim beneath the resolutions.

🤖 Generated with Claude Code