rule: refresh-before-decide extends to working-tree reads — 1008Z empirical anchor caught phantom PR #5128 drift

[AceHack]

Summary

Extends .claude/rules/refresh-before-decide.md with a new section naming the post-fetch working-tree read trap: git fetch origin main updates refs/remotes/origin/main but does NOT promote local HEAD; subsequent file reads through working-tree paths see STALE files, and substrate authored against that state is a phantom catch.

This tick's session IS the empirical anchor: local primary checkout was 11 commits behind origin/main at 10:08Z (local 2774fef5a vs origin 1641da6d2); reading tools/alignment/filter_gate_log.ts + .test.ts via the working tree made PR #5128 (merged 08:22Z, 1h 46min earlier) look unfixed. The refresh-before-decide discipline caught the staleness via git rev-parse HEAD vs git rev-parse origin/main mismatch before any phantom-drift substrate landed.

Three mitigation patterns

1. Isolated worktree off origin/main (default for agent ticks): git worktree add --detach <path> origin/main — composes with agent-worktree-hygiene --detach discipline (never hold main ref in agent worktrees).
2. git show origin/main:<path> — cheapest option for ad-hoc single-file inspection without checkout.
3. ff-promote local HEAD — ONLY when the checkout is the agent's own, never the operator's primary.

Substrate-inventory step (per verify-existing-substrate-before-authoring)

Searched .claude/rules/ + memory/ for: git fetch, FETCH_HEAD, local HEAD stale, ff-only, fast-forward, git show origin/main, stale local. Found existing partial coverage at otto-channels-reference-card.md ID-allocation section ("do NOT use find docs/backlog -name 'B-*.md' on the local worktree...") and refresh-world-model-poll-pr-gate.md ("Prefer origin/main over FETCH_HEAD"). Conclusion: principle exists at NARROW scope (ID-allocation queries / FETCH_HEAD race); the general working-tree-file-read corollary is implicit but not explicit on a cold-boot-loaded rule surface. Disposition: extend (not mint parallel) per the verify-existing-substrate rule — landed inside refresh-before-decide.md so it auto-loads at every Otto cold-boot.

What changed

• .claude/rules/refresh-before-decide.md (+69 lines): new "## `git fetch` updates refs but NOT working-tree files (post-fetch read trap)" section + 3-pattern mitigation + 2026-05-26T10:08Z empirical anchor + citation to existing partial coverage + 5 composes_with entries
• docs/hygiene-history/ticks/2026/05/26/1008Z.md (+151 lines): full tick trace — refresh-found-stale, verify-before-defer composition (9th anchor: 0 stuck procs + 0 peer procs + dotgit-clean → isolated worktree clean), substrate-inventory step, visibility signal

Test plan

• Substrate-inventory step performed per verify-existing-substrate-before-authoring rule (PR rule: verify-existing-substrate-before-authoring — sibling to dep-pin-search-first-authority (3-anchor empirical evidence 2026-05-26) #5131) — citations in rule body + tick shard
• Verify-before-defer composition: git worktree add --detach /private/tmp/zeta-otto-cli-refresh-extend-1008z origin/main → exit=0, HEAD=1641da6d2, ls-tree=61, status=0, no index.lock
• Branch-guard fired pre-commit per zeta-expected-branch.md (git branch --show-current matched ZETA_EXPECTED_BRANCH)
• Post-commit canary per codeql-no-source-on-docs-only-pr-is-broken-commit-canary.md: ls-tree HEAD~1 = HEAD = 61
• Push wrapped in timeout --kill-after=5s 90s per B-0615 discipline
• Stays additive — no existing content removed; existing 28-line rule preserved verbatim

Composes with

• .claude/rules/refresh-before-decide.md — the rule this PR extends
• .claude/rules/verify-existing-substrate-before-authoring.md — PR rule: verify-existing-substrate-before-authoring — sibling to dep-pin-search-first-authority (3-anchor empirical evidence 2026-05-26) #5131; the substrate-inventory rule used here
• .claude/rules/agent-worktree-hygiene-never-hold-main-never-step-on-operator-cleanup-on-pr-merge.md — --detach origin/main discipline
• .claude/rules/refresh-world-model-poll-pr-gate.md — "Prefer origin/main over FETCH_HEAD" + saturation tiers
• .claude/rules/otto-channels-reference-card.md — ID-allocation narrow-scope precedent
• .claude/rules/dep-pin-search-first-authority.md — sibling rule at version-pin scope; same "Otto-defaults-to-plausible-but-unverified" root cause class
• PR #5128 — the fix whose phantom-drift catch this tick prevented

🤖 Generated with Claude Code

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.