Security audit

create-agent

Security checks for vulnerabilities and agentic risk

Overview

The skill's code and runtime instructions are coherent with its stated purpose (creating and registering OpenClaw agents and workspaces) and request only the expected local OpenClaw files and CLI; no unexplained network endpoints or extra credentials are requested — but there are some quality issues and privacy-sensitive behaviors you should review before running it on real data.

This skill appears to do what it says: create workspace files, modify ~/.openclaw/openclaw.json, and register/unregister agents. Before running it on production data: 1) Run operations with the provided --dry-run flags first so you can preview changes. 2) Backup your ~/.openclaw/openclaw.json and any important workspaces manually; the scripts also create backups but you should keep your own. 3) Inspect the scripts (register_agent.py, deregister_agent.py, create_workspace.sh, verify_workspace.sh) — they operate on files under your home dir and will append/modify parent MEMORY.md entries and may restart the gateway; ensure you trust the source. 4) Note privacy: the skill reads/writes MEMORY.md, memory/*.md and org-context.md which may contain PII or company secrets — audit what will be written. 5) There are minor code-quality issues (e.g., in deregister_agent.py the dry-run preview references a variable that is defined later; this is a bug that can cause errors in dry-run flows) — consider reviewing/fixing those before use. 6) Test in a safe environment (a throwaway account or VM) to confirm behavior and validate backups/rollbacks work as documented.

Static analysis

No suspicious patterns detected.