Debugging
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [Prompt Injection] (SAFE): The instructions use strong language to enforce a debugging process but do not attempt to bypass safety filters or jailbreak the agent.
- [Command Execution] (SAFE): The scripts/find-polluter.sh script executes npm test on local files. This is standard developer tooling behavior and is not used for malicious purposes.
- [Data Exposure & Exfiltration] (SAFE): The skill does not access sensitive files or perform any network operations.
- [Indirect Prompt Injection] (SAFE): 1. Ingestion points: scripts/find-polluter.sh reads filenames from the filesystem via the find command. 2. Boundary markers: Filenames are wrapped in double quotes when passed to npm test to prevent shell splitting. 3. Capability inventory: The script executes npm test as a subprocess. 4. Sanitization: Relies on shell quoting; no specific filename sanitization is performed. This is acceptable for a local development utility.
Audit Metadata