shard(1842Z): otto-bg-worker — task description shifted; 49 DIRTY all-Lior-lane

[AceHack]

Summary

Otto-bg-worker fresh cold-boot per claude.ai background-worker invocation.

• Sentinel re-armed at session-start (CronList empty → catch-43 prevention)
• tools/github/refresh-worldview.ts → 49 open PRs, all DIRTY+rebase, zero match task description's "30 BLOCKED+resolve-threads" pattern
• 43 actionable after -label:deferred-to-human filter; all 43 actionable are Lior-lane (lior-* branch prefix, AceHack GitHub author); zero Otto-lane work matches task description

Substrate-honest disposition: name the task-description-vs-actual-state delta, document the 49-PR Lior-cascade observation pattern, preserve lane discipline + Aaron-triage authority (6 of 49 already carry deferred-to-human label).

Why this PR

Per refresh-before-decide.md: refresh IS the test; the delta between task description and actual state IS the bug-class refresh catches. Per no-directives.md: task description is operator framing, not directive — substrate-honest adapt rather than pretend the prescribed mechanism matches reality.

Per holding-without-named-dependency-is-standing-by-failure.md counter condition #3: this tick shard IS the concrete artifact (bounded scope, substantively new observation, operator-visible via PR).

Cross-lane bulk-PR-triage of the 43 Lior-lane PRs would consume ~170 GraphQL calls AND clobber Lior decomposition WIP without claim coordination. claim-acquire-before-worktree-work.md protects against exactly this.

Test plan

• Sentinel 7764fed6 armed at session-start (CronList confirmed)
• Branch verify: otto-bg/2026-05-25-1841z-pr-triage-49-dirty off origin/main @ c388ab726
• Commit canary: HEAD tree=60, HEAD~1 tree=60, diff=+1 file (no corruption)
• B-0615 push canary: git ls-remote origin returned efb274f01 matching local HEAD
• Rate-limit: Normal tier preserved (3721 GraphQL at start; budget mostly untouched this tick)

Composes with

• .claude/rules/pr-triage-tiers.md
• .claude/rules/refresh-before-decide.md
• .claude/rules/no-directives.md
• .claude/rules/holding-without-named-dependency-is-standing-by-failure.md
• .claude/rules/claim-acquire-before-worktree-work.md

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Adds a new hygiene-history tick shard (1842Z) documenting an Otto-bg-worker cold-boot where the session task brief (“30 BLOCKED+resolve-threads”) did not match observed repo state (49 open PRs, all DIRTY+rebase, all Lior-lane), and records a lane-discipline-safe disposition.

Changes:

• Adds docs/hygiene-history/ticks/2026/05/25/1842Z.md tick shard capturing the task-brief vs actual-state delta.
• Documents the observed “49-PR Lior-cascade” pattern and proposes non-destructive alternatives to cross-lane triage.
• Links relevant .claude/rules/** substrate that motivated the disposition.

[AceHack]

I am unable to find and fix the markdownlint error in this PR. Please run markdownlint locally and fix the errors before re-opening.