Contributing

Adding a New Hook

Create a .sh file in hooks/:

#!/usr/bin/env bash
set -euo pipefail
# Read JSON context from stdin
input=$(cat)
# Your logic here
exit 0  # allow
# exit 2  # block

Wire it into .claude/settings.json under the appropriate event (PreToolUse, PostToolUse, etc.):

{
  "hooks": {
    "PreToolUse": [
      { "matcher": "Bash", "hook": "/path/to/hooks/your-hook.sh" }
    ]
  }
}

Document it in hooks/README.md with: purpose, trigger event, and block conditions.

Create a directory in skills/your-skill-name/.
Add a SKILL.md file inside it. This is the instruction file Claude reads when the skill is invoked.
Optionally add helper scripts the skill references.
Test by running /your-skill-name in a Claude Code session.
Document the skill's purpose and usage in skills/README.md.

Use the GitHub issue templates:

Bug report: Include Claude Code version, OS, component, reproduction steps.
Feature request: Describe the feature, which component it affects, and the use case.

Search existing issues before opening a new one.

All bash scripts start with set -euo pipefail.
Hooks read JSON from stdin via input=$(cat) and parse with jq.
Exit codes for hooks: 0 = allow, 2 = block. Any other exit code is treated as an error.
Use jq for JSON parsing, not grep/sed on JSON.
Keep scripts focused: one hook, one responsibility.
Quote all variables: "$var", not $var.