/latch

Hook Recipes

Purpose: Read this when you need a proven hook pattern, recipe ID, or stack-specific recipe combination.

For canonical matcher syntax, exit-code semantics, output schemas, and event contracts, see hook-system.md (§ Matcher Patterns, § Exit Codes, § Output Format, § Event Catalog). This file only references those rules and shows recipe-specific implementations.

Contents

• Recipe catalog
• Security recipes
• Quality recipes
• Context and workflow recipes
• Tech stack sets
• Combining recipes

Recipe Catalog

Security Recipes

S1: File Write Security Validation

Use when writes must avoid system directories, .env, credentials, or traversal.

{
  "PreToolUse": [
    {
      "matcher": "Write|Edit",
      "hooks": [
        {
          "type": "prompt",
          "prompt": "File path: $TOOL_INPUT.file_path. Verify: 1) Not in /etc or system directories 2) Not .env, credentials, or key files 3) Path does not contain .. traversal 4) Not overwriting critical config without reason. Return approve or deny with reason.",
          "timeout": 15
        }
      ]
    }
  ]
}

S2: Bash Command Safety Validation

This is the most common safety recipe and must remain easy to discover.

{
  "PreToolUse": [
    {
      "matcher": "Bash",
      "hooks": [
        {
          "type": "prompt",
          "prompt": "Command: $TOOL_INPUT.command. Check for: 1) rm -rf with broad paths 2) Destructive database commands 3) chmod 777 4) Network operations to unknown hosts 5) Package installs from untrusted sources. Return approve, deny, or ask.",
          "timeout": 15
        }
      ]
    }
  ]
}

S3: Secret Leak Prevention

Keep this example because it is the canonical stderr + exit 2 + permissionDecision: deny pattern.

{
  "PreToolUse": [
    {
      "matcher": "Write|Edit",
      "hooks": [
        {
          "type": "command",
          "command": "bash ~/.claude/hooks/check-secrets.sh",
          "timeout": 10
        }
      ]
    }
  ]
}

#!/bin/bash
set -uo pipefail
input=$(cat)
content=$(echo "$input" | jq -r '.tool_input.content // empty')

if [ -n "$content" ] && echo "$content" | grep -qiE '(api[_-]?key|password|secret|token|private[_-]?key)\s*[:=]\s*["\x27]?[A-Za-z0-9+/]{20,}'; then
  echo '{"hookSpecificOutput":{"permissionDecision":"deny"},"systemMessage":"Potential secret detected in content. Remove secrets before writing."}' >&2
  exit 2
fi

exit 0

S4: MCP Delete Operation Guard

{
  "PreToolUse": [
    {
      "matcher": "mcp__.*__delete.*",
      "hooks": [
        {
          "type": "prompt",
          "prompt": "MCP deletion detected. Verify intent, reversibility, and backups. Return ask unless the deletion is clearly safe.",
          "timeout": 15
        }
      ]
    }
  ]
}

Quality Recipes

Q1: Test Execution Enforcement

Keep this as the default Stop quality gate.

{
  "Stop": [
    {
      "matcher": "*",
      "hooks": [
        {
          "type": "prompt",
          "prompt": "If code was modified, verify that tests were executed. If no tests ran after code changes, block with reason. Otherwise approve.",
          "timeout": 30
        }
      ]
    }
  ]
}

Q2: Auto Lint After Edit

{
  "PostToolUse": [
    {
      "matcher": "Write|Edit",
      "hooks": [
        {
          "type": "command",
          "command": "bash ~/.claude/hooks/auto-lint.sh",
          "timeout": 30
        }
      ]
    }
  ]
}

Use a file-extension switch inside the script so the hook exits quickly for unsupported files.

Q3: Type Check Enforcement

{
  "Stop": [
    {
      "matcher": "*",
      "hooks": [
        {
          "type": "command",
          "command": "bash ~/.claude/hooks/check-types.sh",
          "timeout": 60
        }
      ]
    }
  ]
}

Use when the project has a deterministic type-check command.

Q4: Build Verification

{
  "Stop": [
    {
      "matcher": "*",
      "hooks": [
        {
          "type": "prompt",
          "prompt": "If code was modified, verify that the project build succeeded. If no build ran after code changes, block and request a build step.",
          "timeout": 30
        }
      ]
    }
  ]
}

Context and Workflow Recipes

C1: Project Context Auto-Load

This is the canonical SessionStart context recipe and must stay discoverable.

{
  "SessionStart": [
    {
      "matcher": "*",
      "hooks": [
        {
          "type": "command",
          "command": "bash ~/.claude/hooks/load-project-context.sh",
          "timeout": 10
        }
      ]
    }
  ]
}

Use $CLAUDE_ENV_FILE to persist variables such as PROJECT_TYPE and package-manager hints.

C2: PreCompact Critical Info Preservation

Keep this example because it directly affects context-preserving behavior.

{
  "PreCompact": [
    {
      "matcher": "*",
      "hooks": [
        {
          "type": "command",
          "command": "bash ~/.claude/hooks/preserve-context.sh",
          "timeout": 10
        }
      ]
    }
  ]
}

#!/bin/bash
project_file="$CLAUDE_PROJECT_DIR/.agents/PROJECT.md"

if [ -f "$project_file" ]; then
  content=$(head -50 "$project_file")
  echo "{\"systemMessage\":\"Critical project context (preserve through compaction): $content\"}"
fi

exit 0

W1: Notification Logging

{
  "Notification": [
    {
      "matcher": "*",
      "hooks": [
        {
          "type": "command",
          "command": "bash ~/.claude/hooks/log-notification.sh",
          "timeout": 5
        }
      ]
    }
  ]
}

W2: Temporary Hook via Flag File

Keep this example because it is the main reversible strict-mode pattern.

{
  "PreToolUse": [
    {
      "matcher": "Bash",
      "hooks": [
        {
          "type": "command",
          "command": "bash ~/.claude/hooks/strict-bash-check.sh",
          "timeout": 10
        }
      ]
    }
  ]
}

#!/bin/bash
FLAG_FILE="$CLAUDE_PROJECT_DIR/.enable-strict-validation"

if [ ! -f "$FLAG_FILE" ]; then
  exit 0
fi

input=$(cat)
command=$(echo "$input" | jq -r '.tool_input.command // empty')

if echo "$command" | grep -qE '\brm\b'; then
  echo '{"hookSpecificOutput":{"permissionDecision":"deny"},"systemMessage":"Strict mode: rm commands blocked. Remove .enable-strict-validation to disable."}' >&2
  exit 2
fi

exit 0

Usage:

touch .enable-strict-validation
rm .enable-strict-validation
# Restart Claude Code after changing the flag state

W3: Session Audit Trail

Use paired SessionStart and SessionEnd command hooks to append session start/end lines to a shared log.

Tech Stack Sets

Use stack sets as combinations of recipe IDs plus one or two stack-specific checks instead of copying large JSON blocks.

Combining Recipes

• Merge recipes by concatenating matcher groups under the same event.
• Keep recipe comments or IDs in the config so the purpose stays discoverable.
• Start with 1-2 recipes and add more gradually.
• Too many hooks can slow the workflow or create false positives.

Per-Recipe Behavior Depth

Referenced from SKILL.md -> Subcommand Dispatch.

Per-Recipe behavior depth is documented in the inline sections below. Recipes with a Read First reference follow that reference for full contracts.

configure — Default hook configuration flow

Full SCAN → PROPOSE → IMPLEMENT run. settings.json backup required. Instruct session restart after JSON syntax validation.

debug — Diagnose existing hook issues

Check /hooks → run claude --debug → manual stdin test. Validate timeout, exit code, and stdout/stderr mixing in order.

pretool — PreToolUse specialization

Choose permissionDecision (allow / deny / ask / defer). Block with exit 2. updatedInput must always pair with permissionDecision: allow.

posttool — PostToolUse specialization

exit 0 only (no blocking). Optional context injection via JSON stdout. Can background with async: true.

notification — Notification event hook

Read reference/notification-hook.md first. Branch on message regex via the matcher to route to terminal-notifier / Slack / Discord / desktop sinks. Apply dedup windows, prefer async: true, gate time-based rules with session start time.

sessionstart — SessionStart event hook

Read reference/sessionstart-hook.md first. Fires on session start and after /clear / /compact. Stdout injects into next turn's context (keep <~10K tokens). Offload heavy work to cron + ~/.cache/; the hook itself should be a lazy cat. Use exit 2 only for env validation gates.

security — PreToolUse security guard

Read reference/security-guard-hook.md first. Use permissionDecision: deny for dangerous Bash (rm -rf /, chmod -R 777, force-push to main), sensitive-file Write/Edit (.env, id_rsa, *.pem), secret-regex matches (use updatedInput to redact), and MCP tool ACL via LATCH_BLOCKED_MCP_TOOLS. In CI=true, promote interactive denies to auto-deny.

quarantine — Distribution-side skill/plugin/MCP guard

Read reference/skill-quarantine-hook.md first. Guards the distribution side (vs security's runtime side). Three baselines: SessionStart sha256 drift vs .chain-manifest.json, PreToolUse Bash deny on claudemarketplaces.com installs unless CLAUDE_PLUGIN_INSTALL_ACK=1, SessionStart MCP tool-description pinning to detect rug-pulls. Pairs with the chain agent. Defense against SkillJect, Unicode Tag, Shai-Hulud-class attacks.

claudemd-update — Stop hook CLAUDE.md proposer

Read reference/claude-md-update-proposer.md first. Stop hook extracting "should have known" candidates to .claude/proposals/. Always exit 0 + async: true (advisory only, never trap shutdown). Filters out linter-duplicates, single-anecdote observations, rules better expressed as hooks. Pair with Hone when 3+ proposals accumulate.

skill-telemetry — PreToolUse skill-usage logger

Read reference/skill-usage-telemetry.md first. PreToolUse hook on the Skill matcher appending {ts, skill, session, cwd} JSONL to ${CLAUDE_PLUGIN_DATA:-$HOME/.claude}/telemetry/skill-usage.jsonl. Always async: true + exit 0 (monitoring, never enforcement). No tool_input capture (PII risk). Provides Darwin/Prune/Gauge/Lore with usage signals. Pattern source: Anthropic "Lessons from Building Claude Code".