Skip to content

spec: living-docs regen automation (CI fail-if-stale + make regen-all)#60

Merged
silversurfer562 merged 1 commit into
mainfrom
spec/living-docs-regen-automation
May 24, 2026
Merged

spec: living-docs regen automation (CI fail-if-stale + make regen-all)#60
silversurfer562 merged 1 commit into
mainfrom
spec/living-docs-regen-automation

Conversation

@silversurfer562
Copy link
Copy Markdown
Member

Summary

  • New spec proposing CI fail-if-stale + a make regen-all target as the answer to the recurring manual "regenerate sidecar templates" follow-up commits.
  • Trade-offs vs pre-commit hooks, post-merge bot, and auto-regen-in-CI are captured in the decisions doc.
  • This PR is the spec only — implementation lands in separate PRs once the decision is approved.

Why this approach

Pre-commit blocks local commits and assumes contributor tooling. A bot writing to main needs special access and creates rebase pain. CI fail-if-stale puts the friction in the right place (PR feedback) without giving any automation write access, and the "fix" is one named command.

Test plan

  • Review the decisions doc — are the trade-offs honest, did I miss an option?
  • Sanity-check the "current state" section against your knowledge of the repo
  • Approve or redirect on Approach before any code lands

Open questions in the spec

  1. Exact attune-author maintain invocation for this repo's federated config (a Phase 1 task).
  2. Should CI run the regen check conditionally on changed paths? (Default: yes.)
  3. Runtime budget — combined regen-check must stay reasonable.

🤖 Generated with Claude Code

Almost every code-touching PR needs a manual "regenerate sidecar
templates" follow-up commit. Spec proposes CI-fail-if-stale + a single
`make regen-all` target. Trade-offs vs pre-commit / bot / auto-regen
captured in the decisions doc.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
@silversurfer562 silversurfer562 force-pushed the spec/living-docs-regen-automation branch from d142aa6 to 0683a08 Compare May 24, 2026 02:11
@silversurfer562 silversurfer562 merged commit 31093cb into main May 24, 2026
5 checks passed
silversurfer562 added a commit that referenced this pull request May 25, 2026
attune-rag 0.2.0 shipped 2026-05-25 as the first SemVer-binding
cut — purely additive (new measure_corpus module + alias-file
helpers; no breaking API changes). The current core pin caps at
<0.2, silently locking fresh installs to attune-rag 0.1.23 even
when 0.2.x is available.

Widen the cap to <0.3 to unblock attune-rag 0.2.x installs.

The cap widen is the trigger for promoting the pending
[Unreleased] content to a real release. That content has been
queued on `main` since v0.7.1 (2026-05-22):

  - MCP server Phases 1-5 (PRs #49-#56) — new attune-gui-mcp
    console script with 6 tools (5 read-mostly + gui_set_spec_status
    write); 30+ tests across test_mcp_tools.py and
    test_mcp_integration.py
  - **Status:** parser fix (PR #57)
  - Living-docs regen automation Phase 1 (PRs #60-#63)
  - Home interpreter snapshot (PR #59)
  - Stacked-rebase README note (PR #58)

Bumped 0.7.1 → 0.8.0 (minor) for the new MCP tool surface.

Local smoke (python 3.10, fresh venv):
  pip install -e .[dev]
  → attune-gui 0.8.0, attune-rag 0.2.0
  545/545 sidecar tests pass (1 playwright skip, unrelated).
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