Skip to content

docs(specs): note Phase 2 blockers in living-docs regen automation#62

Merged
silversurfer562 merged 1 commit into
mainfrom
docs/spec-phase2-blocker-note
May 24, 2026
Merged

docs(specs): note Phase 2 blockers in living-docs regen automation#62
silversurfer562 merged 1 commit into
mainfrom
docs/spec-phase2-blocker-note

Conversation

@silversurfer562
Copy link
Copy Markdown
Member

Summary

Phase 2 implementation attempt (after merging #61 for Phase 1) surfaced two foundational blockers. Documenting them in the spec so the next person picking this up doesn't repeat the discovery.

1. attune-author regen vs. staleness check disagree on source_hash. Regen writes a new hash to frontmatter; the immediately-following dry-run still reports the feature stale. Likely cause: regen hashes a budget-truncated source view while staleness hashes the full set. Filed as a spawn-task chip for attune-author.

2. Spec's CI premise conflicts with existing CI policy. The spec proposed make regen-all + git diff --exit-code in CI, which requires ANTHROPIC_API_KEY. But .github/workflows/tests.yml:38-46 has an explicit guard that fails CI if that key is set in the default suite. The --dry-run workaround (no API needed) is blocked by issue #1.

Why this is the right move (not a code PR)

Phase 2 stays parked until attune-author is fixed. Filing a code PR with a half-working CI check would land a permanent red signal. The honest documentation update keeps the spec accurate.

Phase 1 (#61) marked shipped in the spec. Phase 3 (README) still pending and not blocked.

Test plan

  • Read the new "Phase 2 blockers" section — does it accurately capture the issue?
  • Confirm the attune-author chip is the right home for the upstream fix

🤖 Generated with Claude Code

Two foundational issues surfaced during Phase 2 implementation:
1. attune-author regen and dry-run disagree on source_hash — the
   loop never reaches a fixed point.
2. Spec premise (full regen in CI) conflicts with the existing
   ANTHROPIC_API_KEY-must-not-be-set CI guard.

Phase 1 marked shipped. Phase 2 parked pending upstream fix.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
@silversurfer562 silversurfer562 merged commit 502bdbd into main May 24, 2026
5 checks passed
silversurfer562 added a commit to Smart-AI-Memory/attune-author that referenced this pull request May 25, 2026
Phase 2 of attune-gui's living-docs-regen-automation (Smart-AI-Memory/attune-gui#62)
surfaced an attune-author bug: regenerate writes a source_hash that the
immediately-following status check disagrees with. Likely cause: regen
hashes a budget-truncated source view while staleness hashes the full
set.

This decisions.md captures the handoff — repro, code pointers, fix
directions to choose between — so the next session picking this up
starts with the architecture context already laid out. No code fix yet;
that needs Phase 1 (failing test) first.

Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant