gstack/land-and-deploy
James Cruce d2e2618474
fix(land-and-deploy): correct the --auto failure diagnosis in Step 4
Step 4 says `--auto` failing means "repo doesn't have auto-merge enabled".
That is only one of two causes, and in practice it is the rarer one.

`--auto` only QUEUES a merge behind pending required checks. When every
required check has already settled -- or the repo declares no required
status checks at all -- GitHub considers the PR immediately mergeable and
rejects enablePullRequestAutoMerge with "Pull request is in clean status"
or "Pull request is in unstable status". Auto-merge being enabled makes no
difference. A repo with zero required status checks takes the direct path
100% of the time, and so does any repo whose CI finishes before Step 4 runs.

The flow itself is fine -- the direct-merge fallback already handles both
cases. The problem is the recorded reason: an operator reading
MERGE_PATH=direct is told to go check whether auto-merge is enabled, which
sends them to the wrong setting. I lost time to exactly that on a repo where
auto-merge was enabled the whole time and the ruleset simply had no required
checks.

Also adds --squash to the --auto invocation. The fallback squashes, so
without it the two paths merge with different strategies depending on which
one happens to be taken.

Docs-only. Regenerated with `bun run gen:skill-docs` (and verified clean for
--host codex and --host factory); output is idempotent on re-run, so the
Skill Docs Freshness gate stays green.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-02 14:22:30 -04:00
..
SKILL.md fix(land-and-deploy): correct the --auto failure diagnosis in Step 4 2026-08-02 14:22:30 -04:00
SKILL.md.tmpl fix(land-and-deploy): correct the --auto failure diagnosis in Step 4 2026-08-02 14:22:30 -04:00