PRB's .claude directory.
git clone git@github.com:PaulRBerg/dot-claude.git ~/.claude
cd ~/.claude
just install
ccc # Make your first commit with the claude wrapperSee Installation for full setup and Configuration for customization.
- Node.js: Husky/lint-staged automation (
npm install) - Just: command runner for build scripts (
brew install just) - Python 3.13+ and uv: Python package/project manager
git clone git@github.com:PaulRBerg/dot-claude.git ~/.claude
cd ~/.claude
just install # Node deps, Python deps, and CLI utilitiesjust full-check # Run all code checks
just test-hooks # Run hook tests
claude # Run ClaudeAll JSONC files in settings/* merge into settings.json on commit via Husky + lint-staged.
Edit only settings/**/*.jsonc (never settings.json directly). Merging happens on commit, or run just merge-settings
manually.
Settings layout:
basics.jsonc: core config, env vars, status linehooks.jsonc: event hooksplugins.jsonc: enabled pluginspermissions/*.jsonc: permission rules (bash, mcp, read, write, tools)
Global instructions are in CLAUDE.md (communication style, code preferences, scope preservation, git safety, shell
escaping).
Append flags to prompts to trigger behavior:
-s: subagents-c: auto-commit-t: testing-d: debug-n: skip linting
Flags compose, e.g. implement API -s -t -c.
Use just for common tasks like just full-check, just merge-settings, and just test. See justfile for the full
command list.
commands/ contains thin entry points that invoke skills. Commands still matter because they support directory nesting,
which enables namespaced patterns like /yeet:issue-cc and /md-docs:update-readme.
Skills are managed in PaulRBerg/dot-agents and installed via Vercel's
skills CLI. This repo keeps symlinks to ~/.agents/skills/. See dot-agents for
installation guidance.
Examples: commit, code-review, code-polish, vitest, effect-ts, web3-foundry, biome-js, cli-gh, yeet.
agents/ includes specialized subagents:
- tool-finder: discover tools and packages for any task
- docs-finder: find library/framework documentation
Invoke via -s or the Task tool.
Three MCP servers are configured in .mcp.json:
- context7: library docs and code examples
- filesystem: filesystem access and manipulation
- sequential-thinking: chain-of-thought reasoning
Enable or disable them in settings/permissions/mcp.jsonc.
Hooks provide event-driven Claude Code automation. See hooks/README.md.
Active hooks:
- add_plan_frontmatter.py: add YAML frontmatter to plan files (
PostToolUse) - commit_prompts.py: commit logged prompts to zk notebook (
SessionEnd) - sync_global_settings.py: sync global settings across machines (
SessionEnd) - sync_local_settings.py: sync local project settings (
SessionEnd) - log_prompts.py: log conversations to zk notebook (
UserPromptSubmit, optional)
Optional (commented out): ai-flags (prompt flag parsing), ai-notify (desktop notifications).
Two official plugins are enabled via settings/plugins.jsonc:
- claude-md-management: CLAUDE.md auditing and improvement
- plugin-dev: plugin development tools (commands, skills, hooks, agents)
Note
The claude wrapper runs Claude with --dangerously-skip-permissions and auto-loads MCP servers from .mcp.json.
See this issue.
Optional shell utilities from utils.sh:
ccc [args]: streamlined commits via/commit(defaults to--all)cccp: commit and push (feature branches)ccbump [args]: quick release bumping via/bump-releaseclaude [args]: enhanced CLI wrapper with MCP auto-loading
Source them in your shell config:
source ~/.claude/utils.shMIT. See LICENSE.md.