Repository navigation
Bring the documentation up to what the repository actually is - #43
Merged
Merged
Conversation
Three features landed without the front door changing to match: the ChatGPT manifests, the source catalog, and the vertical generator. The README still carried a badge reading "built for Claude Code", which stopped being true the day the second set of manifests shipped, and it is the first thing anyone reads. The badge now says where it runs, and a fourth reports the catalog. The layout tree was missing the marketplace file ChatGPT actually reads, which made the diagram wrong in the one place a ChatGPT user would look. The social card said the same stale thing and now names both tools and carries the catalog figure. The getting-started guide assumed Claude Code throughout — that it was the destination, that installing meant a slash command, and that naming a skill meant a slash form. It now has an install path per tool and says plainly that the slash form is one tool's convention. It also carries the ChatGPT context budget, because that is the one place where installing selectively is not merely tidier but the only thing that works: description text for all sixteen departments comes to roughly six times the budget. Worked situations now say that where a situation turns on an obligation or a threshold, the skill cites the authority that settles it, and point at the catalog. AGENTS.md gained the two rules an agent editing this repository needs and did not have: that sources are references rather than copies and that the license class decides whether one may be quoted, and that verticals emit one way and their output is neither committed nor hand-edited. Its note on the spelling fixer now mentions that it leaves URLs alone, since it did not always. Counts everywhere come from the tree or the catalog rather than from prose, so none of this can drift again on its own. Verified in a browser: the org chart intro, its four figures and the social card all read correctly. All fourteen checks pass. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FaQP2WCqeFYH7s9pi1CJ5u
The generated badge row only ever said what the tree contains. Nothing on the page told a reader whether the checks actually pass, how active the repository is, or who built what — all things a person weighs before installing anything. - A second badge row, fetched at page load rather than generated: checks, stars, contributors, last commit, visitors, and PRs welcome. These cannot go stale because nothing here stores them. - A contributors section crediting the origin of the ChatGPT and Codex support, which came in from outside rather than from this branch. - A star history chart, with a dark variant, since the default canvas is light and the plotted line vanishes against a dark theme. Two shields bases now exist, and the distinction is the reason this is a separate commit from a one-line edit: the static `/badge/` endpoint takes a literal label-message-color triple, while the live endpoints hang off the root. Interpolating the `/badge` base into a `github/...` path yields a URL that renders a picture of its own path instead of the value. Every URL emitted here was fetched and confirmed to return a real value before landing. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FaQP2WCqeFYH7s9pi1CJ5u
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
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
2 participants
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 features landed without the front door changing to match — the ChatGPT manifests, the source catalog, and the vertical generator. You spotted the badge; auditing outward from it found the rest.
README
built for — Claude Coderuns in — Claude Code · ChatGPTcited sources — 181, linking the indexThe layout tree was missing
.agents/plugins/marketplace.json— the file ChatGPT actually reads, so the diagram was wrong in the one place a ChatGPT user would look. Both marketplaces are now shown.Getting started
This was the worst of it. The guide assumed Claude Code end to end: that it was the destination, that installing meant a slash command, and that naming a skill meant a slash form.
It now has an install path per tool, and says plainly that the slash form is one tool's convention rather than how skills are addressed.
It also carries the ChatGPT context budget, which is the one place where installing selectively stops being tidiness and becomes the only thing that works — descriptions for all sixteen departments come to roughly six times the ~8,000-character allowance.
Social card
Said
for Claude Code. Now names both tools and carries the catalog figure, so the image that gets scraped into link previews matches the repository.AGENTS.md
Gained the two rules an agent editing this repo needs and did not have:
licensefield decides whether one may be quoted or only cited — with the instruction to take the more restrictive class when torn.Its note on the spelling fixer now mentions that it leaves URLs alone, since it did not always — that was the bug in #42.
Worked situations
Now say that where a situation turns on an obligation or a threshold, the skill cites the authority that settles it, and point at the catalog.
Checked but already correct
CONTRIBUTING.md(its twelve-row table covers all fourteen checks — the generated-docs row covers three), both marketplace manifests,docs/index.html, and the org chart, which I updated in #41.Verification
Every count comes from the tree or the catalog rather than from prose, so none of this can drift again on its own.
./scripts/check-all.sh— all fourteen green. Driven in a browser: the org chart intro, its four figures, and the social card all read correctly.One thing I cannot change from here: the GitHub repository description still reads "An agent organization for Claude Code…". That is set in repo settings, not in the tree. The
15+ / 125+rounding in it is deliberate and I have left it alone — it is only the tool framing that has gone stale.🤖 Generated with Claude Code
https://claude.ai/code/session_01FaQP2WCqeFYH7s9pi1CJ5u
Generated by Claude Code