fix(tests): pin test-check-rtk-prefix.sh's pre-#113 corpus to a fixed SHA #131

Claude · 2026-09-09T19:41:29Z

Claude commented

2026-09-09 19:41:29 +00:00

Summary

The run-tests pre-push gate was failing on every branch, regardless of what was being pushed — discovered while trying to push unrelated work.

tests/test-check-rtk-prefix.sh's "pre-#113 corpus trips the gate" case reconstructed the historical (pre-sweep) corpus by reading files off the live main ref, asserting the gate should still catch a known drift (gitea-issues/SKILL.md's un-prefixed git remote get-url origin) there. main is the moving integration branch, and the actual fix for that drift (commit ed8c99e) merged back onto main — so the moment it merged, main stopped containing the drift the test exists to catch, and the assertion silently inverted from "this should fail" to "this now passes, incorrectly expected to fail."

Fix

Pin the historical-corpus reconstruction to 598a7c326a4bbfa4a39007eb95d944c4150686e8 — the last commit before ed8c99e where gitea-issues/SKILL.md still had the unprefixed call — instead of the moving main ref. A specific commit SHA is immutable, so this can't drift again the same way.

Also includes a routine apm.lock.yaml refresh to match the deployed install state (the drift AGENTS.md documents as expected from kyberforge's SessionStart hook), needed to clear the apm-audit-ci pre-push gate.

Test plan

bash tests/test-check-rtk-prefix.sh — 23/23 passed
bash tests/run-tests.sh --strict — 26/26 suites passed, 0 skipped, 0 failed
apm audit --ci — 10/10 checks passed

No linked issue — this was discovered as a blocker, not filed ahead of time.

🤖 Generated with Claude Code

https://claude.ai/code/session_01PDj6F7SPXzh3FtPN78dZ88

## Summary The `run-tests` pre-push gate was failing on **every branch**, regardless of what was being pushed — discovered while trying to push unrelated work. `tests/test-check-rtk-prefix.sh`'s "pre-#113 corpus trips the gate" case reconstructed the historical (pre-sweep) corpus by reading files off the live `main` ref, asserting the gate should still catch a known drift (`gitea-issues/SKILL.md`'s un-prefixed `git remote get-url origin`) there. `main` is the moving integration branch, and the actual fix for that drift (commit `ed8c99e`) merged back onto `main` — so the moment it merged, `main` stopped containing the drift the test exists to catch, and the assertion silently inverted from "this should fail" to "this now passes, incorrectly expected to fail." ## Fix Pin the historical-corpus reconstruction to `598a7c326a4bbfa4a39007eb95d944c4150686e8` — the last commit before `ed8c99e` where `gitea-issues/SKILL.md` still had the unprefixed call — instead of the moving `main` ref. A specific commit SHA is immutable, so this can't drift again the same way. Also includes a routine `apm.lock.yaml` refresh to match the deployed install state (the drift AGENTS.md documents as expected from kyberforge's `SessionStart` hook), needed to clear the `apm-audit-ci` pre-push gate. ## Test plan - [x] `bash tests/test-check-rtk-prefix.sh` — 23/23 passed - [x] `bash tests/run-tests.sh --strict` — 26/26 suites passed, 0 skipped, 0 failed - [x] `apm audit --ci` — 10/10 checks passed No linked issue — this was discovered as a blocker, not filed ahead of time. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01PDj6F7SPXzh3FtPN78dZ88

Claude added 2 commits 2026-09-09 19:41:29 +00:00

fix(tests): pin test-check-rtk-prefix.sh's pre-#113 corpus to a fixed SHA 6ba29b696c

The "pre-#113 corpus on main trips the gate" case reconstructed the
historical (pre-sweep) corpus from the live `main` ref. `main` is the
moving integration branch, and the #113 fix (ed8c99e) landed back onto
it — so the moment that fix merged, `main` stopped containing the bare
`git remote get-url origin` drift the case exists to catch, and the
assertion "the gate should fail on this corpus" silently flipped to
false. This blocked `git push` on every branch via the run-tests
pre-push hook, unrelated to whatever was actually being pushed.

Pin to 598a7c3, the last commit before ed8c99e where
gitea-issues/SKILL.md still had the unprefixed call. A specific commit
SHA is immutable, unlike `main`.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PDj6F7SPXzh3FtPN78dZ88

chore: refresh apm.lock.yaml to match the deployed install state 29eefe7f70

The kyberforge SessionStart hook re-resolves dependencies against the
holocron remote on every session start, which routinely leaves
apm.lock.yaml behind the actually-deployed .claude/ content (documented
in AGENTS.md). That mismatch fails apm-audit-ci and apm-pack-check-clean
at the pre-push gate regardless of what's actually being pushed.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PDj6F7SPXzh3FtPN78dZ88

Defame1297 approved these changes 2026-09-09 19:43:01 +00:00

Defame1297 merged commit 058fb5b748 into main

2026-09-09 19:43:06 +00:00

Defame1297 deleted branch fix/rtk-prefix-test-pinned-ref

2026-09-09 19:43:07 +00:00

Defame1297 referenced this issue from a commit

2026-09-09 19:43:08 +00:00

Merge pull request 'fix(tests): pin test-check-rtk-prefix.sh's pre-#113 corpus to a fixed SHA' (#131) from fix/rtk-prefix-test-pinned-ref into main

Sign in to join this conversation.

2 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: Defame1297/holocron#131