Assert this repo pins what it sends its consumers #1402

Merged

coilyco-ops merged 2 commits from aos/claude/ee98-selfconfig into main

2026-08-29 10:50:25 +00:00

coilyco-ops commented

2026-08-29 10:41:02 +00:00

Owner

agentic-os generates .pre-commit-config.yaml for five repos and hand-maintains its own, so the two can disagree indefinitely with nothing to notice. It has happened twice in one lane: #1186 (the generated block passed --force-exclude to typos, this repo's own config passed nothing, so its _typos.toml path excludes were inert and passed only because no flagged word happened to sit in an excluded file) and #1181.

#1186's specific instance already has a test. This adds the general rule it was an instance of: every third-party repo the generated block ships must appear in this repo's own config at the same rev.

Measured on main, all five already agree:

v6.0.0     pre-commit/pre-commit-hooks
v1.7.12    rhysd/actionlint
v12.10.1   forgejo/runner
v0.11.0.1  shellcheck-py/shellcheck-py
v1.48.0    crate-ci/typos

So this lands as a free gate rather than a fix. The two extra entries here (shfmt, ruff) are not in the generated block and are deliberately not asserted.

Verified by injecting a v1.48.0 -> v1.40.0 skew:

E  assert 'v1.40.0' == 'v1.48.0'

I first wrote a second typos-args test and dropped it on finding test_this_repos_own_typos_entry_matches_the_block_it_ships already covers exactly that.

Closes #1199

agentic-os generates `.pre-commit-config.yaml` for five repos and hand-maintains its own, so the two can disagree indefinitely with nothing to notice. It has happened twice in one lane: **#1186** (the generated block passed `--force-exclude` to typos, this repo's own config passed nothing, so its `_typos.toml` path excludes were inert and passed only because no flagged word happened to sit in an excluded file) and **#1181**. #1186's specific instance already has a test. This adds the general rule it was an instance of: **every third-party repo the generated block ships must appear in this repo's own config at the same rev.** Measured on main, all five already agree: ``` v6.0.0 pre-commit/pre-commit-hooks v1.7.12 rhysd/actionlint v12.10.1 forgejo/runner v0.11.0.1 shellcheck-py/shellcheck-py v1.48.0 crate-ci/typos ``` So this lands as a free gate rather than a fix. The two extra entries here (shfmt, ruff) are not in the generated block and are deliberately not asserted. Verified by injecting a `v1.48.0` -> `v1.40.0` skew: ``` E assert 'v1.40.0' == 'v1.48.0' ``` I first wrote a second typos-args test and dropped it on finding `test_this_repos_own_typos_entry_matches_the_block_it_ships` already covers exactly that. Closes #1199

coilyco-ops added 1 commit

2026-08-29 10:41:02 +00:00

test(hooks): assert this repo pins what it sends its consumers

ci / aos-cli-tests (pull_request) Successful in 33s

Details

ci / gate (pull_request) Successful in 59s

Details

c2d720ca7e

agentic-os generates .pre-commit-config.yaml for five repos and
hand-maintains its own, so the two can disagree with nothing to notice.
It has happened twice: #1186 (typos ran without --force-exclude here, so
this repo's own path excludes were inert) and #1181.

#1186's instance already has a test. This is the general rule it was an
instance of: every third-party repo the generated block ships must appear
in this repo's own config at the same rev. The two extra entries here
(shfmt, ruff) are not in the block and are not asserted.

Verified by injecting a v1.48.0 -> v1.40.0 skew, which the test catches.

Closes coilyco-flight-deck/agentic-os#1199

Co-authored-by: Kai Siren <coilysiren@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>
Agent-Role: platform

coilyco-ops added 1 commit

2026-08-29 10:49:14 +00:00