Nginx Proxy Manager

Manage Nginx Proxy Manager (NPM) hosts, certificates, and access lists. Use when the user wants to add a new domain, point a domain to a server/port, enable SSL, or check the status of proxy hosts.

This skill appears to do what it says (manage Nginx Proxy Manager) but has some problematic choices you should consider before installing: - Verify metadata: the registry didn't declare required env vars. SKILL.md and the script require NPM_URL, NPM_EMAIL, and NPM_PASSWORD — ensure you are comfortable providing those credentials. - Prefer token-based or scoped credentials: instead of using a full admin password, create a scoped API token (if NPM supports it) to limit exposure. - Review and/or change TOKEN_FILE: the script writes a token to /root/.npm-token.json (undocumented). If you install/run this, change TOKEN_FILE to a non-root path with strict permissions or remove on-exit persistence to avoid leaving credentials on disk. - Run in a least-privilege context: run the script under a dedicated, limited user account, and ensure the agent runtime cannot expose disk content or network access beyond NPM. - Confirm network targets: the script calls only NPM_URL; ensure that URL is the real admin endpoint you intend to manage and not a malicious redirect. - Ask the author to update metadata and docs: the SKILL registry should declare required env vars and document the token file location and retention policy. If you cannot verify or change these items, treat this skill cautiously (do not supply admin credentials to unknown or untrusted skills).

Type: OpenClaw Skill Name: npm-proxy Version: 1.0.0 The skill is classified as suspicious due to two main indicators. First, the `scripts/npm_client.py` script stores the Nginx Proxy Manager API token in `/root/.npm-token.json`. While this is for session persistence, storing tokens on disk, especially in a potentially privileged directory like `/root/`, is a security risk. Second, the `SKILL.md` file explicitly instructs the AI agent to use `curl` directly for adding new proxy hosts, providing a full example command and payload. Although this command is scoped to the skill's purpose, it represents a direct command execution instruction to the agent, which is a powerful capability that could be misused if the instructions were less constrained.