settings.example.json

{
  "_comment": "Example .claude/settings.json — copy to your project's .claude/settings.json and adjust hook paths.",

  "hooks": {
    "PreToolUse": [
      {
        "_comment": "Block dangerous shell quoting patterns (backtick-in-double-quote, bad f-strings, set slicing).",
        "matcher": "Bash",
        "hooks": [
          {
            "type": "command",
            "command": "python3 \"$CLAUDE_PROJECT_DIR/hooks/pre_shell_safety.py\""
          }
        ]
      },
      {
        "_comment": "Block destructive SQL (DROP, TRUNCATE, DELETE FROM, INSERT INTO SELECT) in Bash commands and file writes.",
        "matcher": "Bash|Write|Edit",
        "hooks": [
          {
            "type": "command",
            "command": "python3 \"$CLAUDE_PROJECT_DIR/hooks/pre_destructive_writes.py\""
          }
        ]
      },
      {
        "_comment": "Enforce proxy routing when HTTPS_PROXY / HTTP_PROXY is set (or AGENT_PROXY_ENFORCE=1).",
        "matcher": "Bash|Write|Edit",
        "hooks": [
          {
            "type": "command",
            "command": "python3 \"$CLAUDE_PROJECT_DIR/hooks/pre_proxy_enforcement.py\""
          }
        ]
      }
    ],

    "PostToolUse": [
      {
        "_comment": "Warn when Bash output is large enough to thrash the context window.",
        "matcher": "Bash",
        "hooks": [
          {
            "type": "command",
            "command": "python3 \"$CLAUDE_PROJECT_DIR/hooks/post_output_size.py\""
          }
        ]
      },
      {
        "_comment": "Scan Bash output for prompt injection signals (instruction overrides, persona shifts, exfiltration probes).",
        "matcher": "Bash",
        "hooks": [
          {
            "type": "command",
            "command": "python3 \"$CLAUDE_PROJECT_DIR/hooks/post_injection_scan.py\""
          }
        ]
      }
    ]
  }
}