Back to skill
Skillv1.4.0
ClawScan security
Cs Qweather Jwtgen · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
ReviewApr 9, 2026, 5:02 PM
- Verdict
- Review
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill appears to implement exactly the JWT generation it claims, but there are coherence issues (undisclosed required environment variables and automatic loading of ~/.openclaw/.env and a local private key) that you should review before installing.
- Guidance
- This skill's code does what it says (generates EdDSA JWTs for 和风天气) and writes the token to ~/.myjwtkey/last-token.dat, but there are a few red flags to consider before installing: - Confirm required inputs: The registry metadata lists no required env vars, but the SKILL.md and script require QWEATHER_SUB and QWEATHER_KID. Make sure you know and trust where you'll provide these values. - Inspect ~/.openclaw/.env: The script attempts to auto-load that file (dotenv with override=True). That may overwrite process env vars and can contain other secrets. Review its contents before allowing the skill to run. - Verify private key location and permissions: The script expects your Ed25519 private key at ~/.myjwtkey/ed25519-private.pem and will read it. Ensure the file is the key you intend to use and has safe permissions (the script sets token file to 600 but you should confirm the private key file is protected). - Run in an isolated environment first: If you are unsure, run the script in a test account or sandbox so you can verify behavior (no network calls are present, but it will read/write local files). - Dependency hygiene: Install pyjwt from a trusted source and avoid untrusted forks. The script optionally uses python-dotenv; if you install that, ensure it's from PyPI. If you require more assurance, ask the skill author to update registry metadata to declare QWEATHER_SUB and QWEATHER_KID as required and to document explicit file path usage, or request a signed/verified source URL and owner provenance before trusting the skill.
Review Dimensions
- Purpose & Capability
- noteThe code and SKILL.md match the described purpose (EdDSA JWT generation for 和风天气). However the registry metadata lists no required environment variables or credentials while SKILL.md and the script require QWEATHER_SUB and QWEATHER_KID and a local private key path; this mismatch is an integrity concern (the skill will fail or behave unexpectedly if required inputs are not provided).
- Instruction Scope
- okThe runtime instructions and the script stay within the stated task: read a local private key, read environment variables, generate a JWT with pyjwt, write the token to ~/.myjwtkey/last-token.dat, and log to /tmp/cslog. The script does not attempt network calls or other unrelated data collection.
- Install Mechanism
- okNo install spec; this is instruction-only with a Python script. Dependencies are minimal (pyjwt, optional python-dotenv). There is no downloading of remote archives or execution of arbitrary installers.
- Credentials
- concernThe script requires QWEATHER_SUB and QWEATHER_KID but the registry metadata declared none; SKILL.md documents these but the platform-level metadata omission is an inconsistency. The script also auto-loads ~/.openclaw/.env (dotenv.load_dotenv(..., override=True)) which reads a user-specific env file and will override environment variables — this is sensitive because that file may contain other secrets. The skill reads a private key at ~/.myjwtkey/ed25519-private.pem and writes ~/.myjwtkey/last-token.dat; these file accesses are expected for the task but are not declared in the registry metadata.
- Persistence & Privilege
- okThe skill does not request always:true and does not modify other skills or system-wide settings. It persists the generated token locally (~/.myjwtkey/last-token.dat) and writes logs to /tmp/cslog; this is reasonable for its function, but users should be aware of local file writes.