Curated collection of ready-to-use Claude Code hook recipes for safety, automation, formatting, logging, and context injection. Use when configuring hooks, setting up safety guards, automating workflows, or when user mentions 'hooks', 'PreToolUse', 'PostToolUse', 'guard', 'automation', 'claude hooks'.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Curated collection of ready-to-use Claude Code hook recipes for safety, automation, formatting, logging, and context injection. Use when configuring hooks, setting up safety guards, automating workflows, or when user mentions 'hooks', 'PreToolUse', 'PostToolUse', 'guard', 'automation', 'claude hooks'.
type
skill
category
patterns
status
stable
origin
taches-cc-resources
modified
true
first_seen
"2026-02-07T00:00:00.000Z"
first_path
examples/hook-recipes/SKILL.md
superseded_by
null
Hook Recipes
A curated collection of ready-to-use Claude Code hook configurations. Each recipe is a complete JSON snippet you can add to your .claude/settings.json or project-level .claude/settings.local.json.
Hook Types Reference
Hook Type
Fires When
Common Use
PreToolUse
Before any tool executes
Block dangerous commands, inject context
PostToolUse
After any tool executes
Log actions, format output, validate results
Notification
Claude wants to notify the user
Custom notification routing
Stop
Claude is about to stop responding
Enforce checklists, add summaries
Tool Names for Matching
Tool Name
What It Does
Bash
Shell command execution
Read
File reading
Write
File writing (full file)
Edit
File editing (partial)
Glob
File pattern matching
Grep
Content searching
WebFetch
HTTP requests
Safety Hooks
Block Destructive Git Commands
Prevents accidental force pushes, hard resets, and branch deletions.
What it does: Intercepts Bash tool calls and blocks commands matching force push, hard reset, clean, and restore patterns. The hook exits 0 (block) if a destructive pattern is found, exit non-zero (allow) otherwise.
When to use: Always. This is a foundational safety hook for any project.
Block Database Destructive Operations
Prevents DROP TABLE, TRUNCATE, and DELETE without WHERE.
{"hooks":{"PreToolUse":[{"matcher":"Bash","hooks":[{"type":"intercept","command":"bash -c '[[ \"$CLAUDE_TOOL_INPUT\" =~ (npm\\s+audit\\s+fix\\s+--force|npm\\s+install\\s+--force|--legacy-peer-deps) ]]'","timeout":5000,"description":"Block force install and audit fix --force"}]}]}}
What it does: Blocks npm audit fix --force (can introduce breaking changes), npm install --force (bypasses version checks), and --legacy-peer-deps (ignores peer dependency conflicts).
When to use: Any Node.js project where dependency integrity matters.
Formatting Hooks
Auto-Format on File Write
Runs Prettier on files after Claude writes or edits them.
{"hooks":{"PostToolUse":[{"matcher":"Write|Edit","hooks":[{"type":"command","command":"bash -c 'FILE=$(echo \"$CLAUDE_TOOL_INPUT\" | jq -r \".file_path // .filePath // empty\"); if [[ -n \"$FILE\" && \"$FILE\" =~ \\.(ts|tsx|js|jsx|json|css|md)$ ]]; then npx prettier --write \"$FILE\" 2>/dev/null; fi'","timeout":10000,"description":"Auto-format files with Prettier after write/edit"}]}]}}
What it does: After every Write or Edit, extracts the file path and runs Prettier if the file matches supported extensions. Fails silently if Prettier is not installed.
When to use: Projects with Prettier configured. Ensures Claude's output matches project formatting.
Auto-Lint on File Write
Runs ESLint fix on written files.
{"hooks":{"PostToolUse":[{"matcher":"Write|Edit","hooks":[{"type":"command","command":"bash -c 'FILE=$(echo \"$CLAUDE_TOOL_INPUT\" | jq -r \".file_path // .filePath // empty\"); if [[ -n \"$FILE\" && \"$FILE\" =~ \\.(ts|tsx|js|jsx)$ ]]; then npx eslint --fix \"$FILE\" 2>/dev/null; fi'","timeout":15000,"description":"Auto-lint files with ESLint after write/edit"}]}]}}
When to use: Projects with ESLint configured. Pairs well with the Prettier hook above.
Logging Hooks
Audit Log of All Tool Usage
Logs every tool invocation to a file for review.
{"hooks":{"PreToolUse":[{"matcher":".*","hooks":[{"type":"command","command":"bash -c 'echo \"$(date -u +\"%Y-%m-%dT%H:%M:%SZ\") [PRE] $CLAUDE_TOOL_NAME: $(echo $CLAUDE_TOOL_INPUT | head -c 200)\" >> /tmp/claude-audit.log'","timeout":3000,"description":"Log all tool usage to audit file"}]}],"PostToolUse":[{"matcher":".*","hooks":[{"type":"command","command":"bash -c 'echo \"$(date -u +\"%Y-%m-%dT%H:%M:%SZ\") [POST] $CLAUDE_TOOL_NAME: exit=$CLAUDE_TOOL_EXIT_CODE\" >> /tmp/claude-audit.log'","timeout":3000,"description":"Log tool completion to audit file"}]}]}}
What it does: Creates a timestamped audit trail of every tool call (before and after) in /tmp/claude-audit.log. Truncates input to 200 chars to keep logs manageable.
When to use: During development and debugging of hook configurations. Also useful for compliance and audit trails.
When to use: When you want to track what Claude changed without the noise of read operations.
Context Injection Hooks
Inject Project Context on Session Start
Automatically load project context when a Claude session begins.
{"hooks":{"PreToolUse":[{"matcher":"Bash|Read|Write|Edit","hooks":[{"type":"command","command":"bash -c 'if [ ! -f /tmp/.claude-context-loaded ]; then cat .claude/project-context.md 2>/dev/null; touch /tmp/.claude-context-loaded; fi'","timeout":5000,"description":"Inject project context on first tool use"}]}]}}
What it does: On the first tool invocation of a session, outputs the contents of a project context file. Uses a flag file to avoid repeating. Output is included in Claude's context.
When to use: Projects with important conventions or context that Claude should always know about.
Inject Recent Git Context
Provides recent commit context before writes.
{"hooks":{"PreToolUse":[{"matcher":"Write|Edit","hooks":[{"type":"command","command":"bash -c 'FILE=$(echo \"$CLAUDE_TOOL_INPUT\" | jq -r \".file_path // .filePath // empty\" 2>/dev/null); if [[ -n \"$FILE\" && -f \"$FILE\" ]]; then echo \"Recent changes to $FILE:\"; git log --oneline -3 -- \"$FILE\" 2>/dev/null; fi'","timeout":5000,"description":"Show recent git history for files being modified"}]}]}}
When to use: When you want Claude to be aware of recent changes to files it's about to modify.
Stop Criteria Hooks
Enforce Checklist Before Stopping
Reminds Claude to verify its work before finishing.
{"hooks":{"Stop":[{"hooks":[{"type":"command","command":"bash -c 'echo \"Before stopping, verify: 1) All tests pass 2) No lint errors 3) Changes committed 4) Summary provided\"'","timeout":3000,"description":"Remind Claude to verify work before stopping"}]}]}}
What it does: When Claude is about to stop, injects a checklist reminder into the context. Claude will see this and either confirm or address missing items.
IMPORTANT WARNING about Stop hooks: A Stop hook that causes Claude to continue, combined with conditions that always re-trigger the Stop hook, creates an infinite loop. Always ensure Stop hooks have a termination condition or are purely informational (exit non-zero to allow stopping).
When to use: When you want consistent completion behavior.
Enforce Test Run Before Stopping
Blocks Claude from stopping until tests pass.
{"hooks":{"Stop":[{"hooks":[{"type":"intercept","command":"bash -c 'if [ -f /tmp/.claude-tests-passed ]; then exit 1; fi; echo \"Tests have not been verified. Run tests before completing.\"; exit 0'","timeout":5000,"description":"Require tests to pass before stopping"}]}]}}
What it does: Blocks Claude from stopping unless a flag file exists. Claude must run tests and create the flag (e.g., npm test && touch /tmp/.claude-tests-passed) before it can complete.
IMPORTANT: This hook has a built-in escape via the flag file. Without an escape condition, Stop intercept hooks create infinite loops.
When to use: Strict TDD workflows where tests must always be green before session ends.
Notification Hooks
Desktop Notification on Completion
Sends a system notification when Claude finishes a task.
{"hooks":{"Stop":[{"hooks":[{"type":"command","command":"bash -c 'notify-send \"Claude Code\" \"Task completed\" 2>/dev/null || osascript -e \"display notification \\\"Task completed\\\" with title \\\"Claude Code\\\"\" 2>/dev/null || true'","timeout":5000,"description":"Send desktop notification on task completion"}]}]}}
What it does: Sends a desktop notification using notify-send (Linux) or osascript (macOS). Falls back silently if neither is available.
When to use: When running long tasks and you want to be notified upon completion.
Composing Hooks
You can combine multiple hooks on the same event. They execute in order.
