Skip to content

Add CLAUDE.md for repository navigation#29363

Open
xml wants to merge 2 commits intoanthropics:mainfrom
xml:add-claude-md
Open

Add CLAUDE.md for repository navigation#29363
xml wants to merge 2 commits intoanthropics:mainfrom
xml:add-claude-md

Conversation

@xml
Copy link
Copy Markdown

@xml xml commented Feb 27, 2026

Summary

  • Add a CLAUDE.md that orients contributors (human and AI) to this repository
  • Lead with the single most important fact: this repo does not contain the Claude Code CLI source code — it hosts the plugin ecosystem, changelog, issue tracker, and security policy
  • Document the plugin architecture (four composable abstractions), standard plugin structure, key conventions, and the scripts/settings in the repo

Motivation

Shouldn't the repo called claude-code really include a CLAUDE.md?! You know you
want to! And besides — since humans never bother to read docs and source, let's
make it easy for Claude to navigate for them.

The repo has great plugin READMEs, but nothing tying it all together at the top
level. The two existing SessionStart hooks inject communication style instructions,
not repository context — so nothing currently fills the role a CLAUDE.md would.

Prior art

PR #20824 is a draft CLAUDE.md from January 25 (zero reviews, zero comments). It
takes a different approach — 261 lines oriented toward AI assistant best practices,
including installation methods, platform support, and model selection that duplicate
the README.

This PR is 124 lines, leads with what the repo actually contains, and avoids
duplicating content from existing READMEs.

Test plan

  • Read the CLAUDE.md and confirm it is accurate
  • Confirm it does not duplicate information from README.md or plugins/README.md
  • Verify the plugin count (13) and script listing (6 files) match the repo

🤖 Generated collaboratively by Claude Code and this human.

xml and others added 2 commits February 21, 2026 15:06
The plugin repo had no CLAUDE.md or SessionStart hook providing
repo-level context. The two existing SessionStart hooks (explanatory
and learning output styles) only inject communication style
instructions, not repository structure or conventions.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant