Back to MCP & Tools
Security & Quality Featured
Hooks
Deterministic control over agent behavior — run a formatter after every edit or block dangerous commands before they execute.
Claude Code
Overview
Hooks are shell commands that execute automatically at defined points of the agent loop: before a tool call (PreToolUse), after it (PostToolUse), when a response finishes (Stop), and more. The key difference from written instructions: a hook always runs — it is not a suggestion the model can forget.
The most common uses: auto-running Prettier or ESLint on every file the agent edits, blocking dangerous Bash commands (rm -rf and friends) before execution, logging every operation for audits, or sending a notification when a long task completes.
A PreToolUse hook can return an exit code that blocks execution — making hooks a real security layer: your policies are enforced at the system level, not by model goodwill.
The most common uses: auto-running Prettier or ESLint on every file the agent edits, blocking dangerous Bash commands (rm -rf and friends) before execution, logging every operation for audits, or sending a notification when a long task completes.
A PreToolUse hook can return an exit code that blocks execution — making hooks a real security layer: your policies are enforced at the system level, not by model goodwill.
Who Is This For?
- Enforcing formatting and linting automatically on every edit
- Blocking dangerous commands or writes to sensitive files
- Keeping an audit log of all agent operations
- Notifications when long-running tasks complete
Installation
Setup for Claude Code
Add a hooks block to .claude/settings.json — e.g. run Prettier after every file edit (PostToolUse) or block dangerous Bash commands (PreToolUse).Configuration
{"hooks":{"PostToolUse":[{"matcher":"Edit|Write","hooks":[{"type":"command","command":"npx prettier --write "$CLAUDE_FILE_PATHS""}]}]}}Explore AI Tools
Discover the best AI tools that complement your skills
Read AI & Design Articles
Tips and trends in the world of design and AI
Related Servers
Security & Quality
Snyk Security Scan
Detect vulnerabilities in your dependencies and application code. Get actionable remediation advice and automatic fix pull requests.
Claude Code Codex Copilot
Security & Quality SonarQube Code Quality
Run continuous code quality and security analysis to catch bugs, code smells, and vulnerabilities before they reach production.
Claude Code Codex Copilot
Security & Quality OWASP ZAP Security Testing
Perform automated web application security testing to find common vulnerabilities like XSS, injection flaws, and misconfigurations.
Claude Code Codex