Clawdbot Security Suite

**Quick judgment:** Clawdbot Security Suite is a strong, specialized security gate for Clawdbot agents that execute shell commands, access external URLs, or handle untrusted content. It is well-suited as a default security layer in any agent that can run a shell, but it should be treated as a robust filter—not a complete security solution. **What it does well** - Adds a **pre-execution security check** for bash commands, URLs, and file paths, making it easy to enforce “validate before execute/fetch” patterns. - Detects **command injection**, **SSRF**, **path traversal**, **API key leaks**, and basic **prompt injection** via pattern-based rules. - Runs **locally only**, with optional local logging and no external telemetry, which is important for privacy-sensitive environments. - Provides **monitoring & audit tools** (`events`, `threats`, `stats`) and a pattern-update mechanism (`update-patterns`) for evolving threat intelligence. **Key risks & limitations** - Detection is **pattern-based**, so sophisticated or novel attacks, obfuscated payloads, or context-dependent exploits can still bypass it. - There is a meaningful risk of **false positives** (legitimate commands, URLs, or file operations being blocked), especially in stricter configurations. - Security depends on **correct integration**: commands or tools not routed through `security.sh` remain unprotected. - Requires a **shell-capable environment** and manual installation/configuration; less suitable for constrained or non-Unix runtimes. **Recommended scenarios** - Production-like Clawdbot agents that: - Execute shell commands influenced by user input. - Perform `web_fetch` or HTTP requests to user-specified URLs. - Read/write user-controlled file paths or process arbitrary external content. - Any agent where you want: - A **fail-closed guardrail** before dangerous tools (e.g., `bash`, file system, HTTP clients). - **Security event visibility** and logs for incident analysis. In practice, this skill is best used as a **standard security layer** around all potentially dangerous tools in Clawdbot, combined with broader security practices (principle of least privilege, sandboxing, rate limiting, and regular pattern updates).