Repository navigation
feat(speckit-ext): file catalog updates as an issue, not a PR - #587
Merged
Merged
Conversation
Our docs told us to update the community catalog with a PR against extensions/catalog.community.json. The upstream guide forbids exactly that, and github/spec-kit#3937 was closed unmerged with "Updates to extensions must use the extension submission issue template." Both first listings and version updates go through the same submission issue. Adds /submit-catalog-update, which renders the issue body from the LIVE upstream issue-form template rather than hardcoded headings — a heading that drifts from theirs breaks the maintainers' validation, and that is the whole failure this guards. Verified its rendered heading set is identical to our accepted submission #2926, including the "(optional)" suffixes. It merges onto the live catalog entry instead of rebuilding it, so verified/downloads/stars/created_at survive; keeps the version-pinned download URL; and passes no --label, since we hold only pull upstream and --label would fail the create outright. Also re-syncs the manifest tags. They were spec-driven-development, tracking, companion — one of them the extension's own name, none covering living specs, drift, or the composable command model. The catalog had separately drifted to a different set of six, so the two now come from one place. Corrects the docs that produced the rejected PR, and the README's command count (its own tables list 18, the prose said seventeen).
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Our release docs told us to update the spec-kit community catalog by opening a PR against
extensions/catalog.community.json. The upstream publishing guide forbids exactly that — "Do not open a pull request directly to editextensions/catalog.community.json" — and github/spec-kit#3937 was closed unmerged: "Updates to extensions must use the extension submission issue template."So the catalog entry is still pinned at 0.11.0 while we ship 0.20.2, and following our own docs is what wasted the round trip. This corrects the docs and adds the tool that does it properly.
The interesting part: heading drift
GitHub serializes an issue form as
### <label>/ blank / value, and the maintainers' validation parses that. Filing withgh issue create --body-filemeans hand-building that markdown — so hardcoded headings that drift from the upstream template would silently break validation.render_submission.pytherefore fetches.github/ISSUE_TEMPLATE/extension_submission.ymlfrom github/spec-kit at run time and renders from its live labels. If it can't fetch, it aborts rather than falling back to a remembered shape — that fallback is precisely the failure it exists to prevent.Verified against our own accepted submission #2926: the rendered heading set is identical, including the
(optional)suffixes (### Homepage (optional),### Number of Hooks (optional)) that would be easy to guess wrong.Other things the tool gets right
verified,downloads,stars, andcreated_atthat exist nowhere inextension.yml; rebuilding drops them and reads as a destructive edit.--label. We hold onlypullon github/spec-kit, sogh issue create --labelerrors and nothing gets filed. Maintainers applyextension-submissionat triage.companion-latest— the catalog requires declared version to equal downloaded bits.categoryanddescriptionare explicit overrides, not silent defaults, because a reclassification or a copy change is editorial and shouldn't ride along with a version bump.|block scalars, so it carries a small purpose-built parser.Manifest tags re-synced
They were
spec-driven-development, tracking, companion— one of them the extension's own name, none covering living specs, drift, or the composable command model added since. Nowvscode, progress, living-specs, drift, hooks.Chosen against all 155 catalog entries:
living-specshas 0 other users so we coin and own it, whiledrifthas 5 and carries the actual search traffic for "my specs went stale". Droppedstatus/resume(near-duplicates ofprogress, and 2 others already rank forresume) and the vagueconfigurable/extensible. The catalog had separately drifted to a different set of six — the two now come from one place.Quality checklist
companion-latestoccurrences;verified/downloads/stars/created_atall preservedpublish-speckit-ext.mdstep 10-11 andpublishing.mdsteps 9-10; the stale paste block (pinned at0.11.0,>=0.8.5, 6 commands, and acompanion-latestURL contradicting the pinning rule) is replaced with a pointer, since a paste-ready copy only rotsspeckit-extension/CHANGELOG.mdunder## [Unreleased]Gaps / how to verify
should list all 23 fields with ids, types, and live labels.
The upstream submission issue for 0.20.2 is not filed by this PR — it's a separate outward action, held for explicit approval.
Two things deliberately left alone: by-name install stays unavailable (the community catalog is discovery-only, independent of this), and
extension.yml's per-command descriptions for specify/plan/tasks are stale relative to behavior — real, but a separate cleanup that doesn't affect the form.🤖 Generated with Claude Code