CLAUDE.md

This file stacks on top of the workspace root at C:\Code\GitHub\:

• Root CLAUDE.md -- voice, rules, routing map, references, skills, slash commands, conventions.
• Root MEMORY.md -- live facts across repos.
• Root STATUS.md -- live PR/CI/security dashboard.
• .claude/resources/ -- deep reference for collaboration, workflow, git, OSS, debugging, voice.

Read those first. The guidance below only adds repo-specific context -- it does not override anything in the root.

Claude Code Recipes - Project Config

Project

This is a recipe collection repository for Claude Code. It contains copy-paste commands, subagents, hooks, skills, MCP configs, workflows, and CLAUDE.md templates. There is no application code - only Markdown files, shell scripts, and JSON configs.

Structure

commands/          # Slash commands (.md) for .claude/commands/
subagents/         # Agent definitions (.md) for .claude/agents/
hooks/             # Shell scripts (.sh) and settings examples for hooks
skills/            # SKILL.md files organized in subdirectories
mcp-configs/       # MCP server configuration (.json) for .mcp.json
workflows/         # End-to-end workflow guides (.md)
claude-md/         # CLAUDE.md templates for different project types
.claude-plugin/    # marketplace.json, generated by scripts/build_marketplace.py
scripts/           # Validator, hook behavior tests, marketplace generator, postgres MCP probe

The repo is also a plugin marketplace with five plugins (recipe-commands, recipe-agents, recipe-skills, recipe-hooks, recipe-guards). Each plugin's source is one recipe directory and lists its files by name, so the directory READMEs never load. Never hand-edit .claude-plugin/marketplace.json: change the recipes or scripts/build_marketplace.py, then run python scripts/build_marketplace.py. Its version comes from the newest CHANGELOG.md heading.

Gates: bash scripts/validate-recipes.sh, bash scripts/test-hooks.sh, shellcheck -S style hooks/*.sh scripts/*.sh, claude plugin validate --strict .

Conventions

• Every recipe must work out of the box - no setup beyond copying the file
• Markdown files use standard GitHub-Flavored Markdown
• Shell scripts use #!/usr/bin/env bash and set -euo pipefail
• JSON files include _comment fields for documentation
• Each directory has a README.md index explaining its contents
• File names use kebab-case: code-review.md, pre-commit-lint.sh
• Skills use SKILL.md (uppercase) inside a kebab-case directory

Writing Style

• Be specific and actionable - concrete rules over vague principles
• Include examples (good and bad) in skills and commands
• Keep recipes self-contained - don't require reading other files
• Use imperative mood in instructions: "Run the tests" not "You should run the tests"
• Recipes ship emoji-free; enforce on contributor PRs too

Do NOT

• Add application code - this is a recipe collection, not a framework
• Create deeply nested directory structures - keep it flat
• Write recipes that require specific API keys or paid services to function
• Include project-specific conventions - recipes must be generic