fix(ci): stop the seal check and the auto-tag blocking releases - #52
Merged
Merged
Conversation
The Pages deploy has failed since #50 merged: its seal-and-verify step fails the build on any claim that is not VALID, and `use/anti-slop.md` was edited in #50 without resealing. The v0.14.0 release added a second stale page, `use/claude-code-recipe.md`, whose sample `claude plugin list` output carries the version. Reseal both with the house key so the site can publish the release. Note the asymmetry this exposes: tests/validate-innsigle.sh only warns on STALE (so PR checks stay green without 1Password), while pages.yml still fails on it. A stale seal therefore passes review and breaks the deploy. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The auto-tag workflow has never tagged a release. It interpolated the PR
title into the shell as `TITLE=${{ toJSON(...) }}`, so bash consumed the
JSON quotes and `json.loads` got a bare string:
json.decoder.JSONDecodeError: Expecting value: line 1 column 1
Both runs to date failed this way — v0.13.3 (which is why that version
was never tagged) and v0.14.0, whose tag was pushed by hand and verified
by release-version-guard.yml instead.
Pass the title as an environment variable and read it directly. This
also keeps a title with shell metacharacters out of the script text.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The deploy called `innsigle verify --all` raw, which fails on STALE. tests/validate-innsigle.sh — the gate the website workflow runs, and the one contributors see on a pull request — treats an unsigned or stale page as a warning and fails only on a broken or wrong-key signature, because sealing a curated page needs the human key from 1Password and CI does not have it. So a page could pass review and then break the deploy. It did twice: #50 edited use/anti-slop.md without resealing, and v0.14.0 bumped the version string in the recipe page. Neither is a signature failure; a stale seal simply does not render. Run the gate here instead of a second, stricter policy nobody reviews. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
easel
changed the title
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three release-plumbing fixes found while cutting v0.14.0.
The seal check blocked the deploy on content that passed review
pages.ymlcalledinnsigle verify --allraw, which fails on STALE.tests/validate-innsigle.sh— the gate the website workflow runs, and the onea contributor sees on a pull request — treats an unsigned or stale page as a
warning and fails only on a broken or wrong-key signature. #42 made that
choice deliberately: sealing a curated page needs the human key from 1Password,
which CI does not have.
So a page could pass review and then break the deploy, and it did twice: #50
edited
use/anti-slop.mdwithout resealing, and v0.14.0 bumped the versionstring in
use/claude-code-recipe.md. The Pages deploy has failed on everypush to
mainsince #50 merged. Neither case is a signature failure — a staleseal costs that page its rendered seal and nothing else.
The deploy now runs the same gate, so one file defines the policy.
Both pages are also resealed here, which is worth doing on its own but is no
longer what unblocks the deploy.
The release auto-tag has never tagged anything
release-tag.ymlinterpolated the PR title into the shell asTITLE=${{ toJSON(...) }}. Bash consumed the JSON quotes, sojson.loadsreceived a bare string:
Both runs to date failed this way: v0.13.3 (which is why that version was
never tagged) and v0.14.0. The v0.14.0 tag was pushed by hand at the merge
commit
33a43087and verified byrelease-version-guard.yml, which passed.The title now reaches the script as an environment variable and is read
directly, which also keeps a title containing shell metacharacters out of the
script text.
🤖 Generated with Claude Code