PostToolUse: Enrich After It Happens

What PostToolUse Can Do

• **Auto-cache expensive results** — every time the AI calls the Figma screenshot tool, a PostToolUse hook saves the image to disk. Next time you need it, it's already cached — no API call needed.
• **Validate file edits** — after any Edit tool call to a plugin file (YAML, JSON, markdown), a hook checks the file is still valid. Catches YAML syntax errors, missing required fields, and accidental deletions.
• **Log subagent completions** — after the Agent tool completes, a hook logs the agent name, model used, and duration. Useful for tracking cost and identifying slow agents.

Hooks to Build First

1. 1 **Cannot block, only observe** — the action already happened. If you need to prevent something, use PreToolUse instead.
2. 2 **Gets tool result as context** — your script can inspect the output and take different actions based on success or failure.
3. 3 Create a PostToolUse hook that logs every Bash command to a file — after a session, review what commands the AI ran
4. 4 Add a validation hook for Edit calls on config files (YAML, JSON) — catch corruption immediately instead of discovering it later
5. 5 Use the matcher to target specific tools — `Edit` for file validation, `mcp__*figma*` for Figma caching