metaculus
Pass
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION] (SAFE): The code does not attempt to access sensitive system files (e.g., SSH keys, AWS credentials) or environment variables. All data handled is public forecasting information.
- [COMMAND_EXECUTION] (SAFE): No use of
eval(),exec(), or subprocess modules was detected. The code uses a structured feed object for data retrieval. - [REMOTE_CODE_EXECUTION] (SAFE): There are no patterns involving the download and execution of remote scripts or unverified packages.
- [PROMPT_INJECTION] (LOW): As an indirect injection surface, the skill fetches external content (question descriptions and titles) from Metaculus. However, the skill lacks high-privilege capabilities such as file-writing or network POST requests to non-whitelisted domains, rendering this vector low-risk. Evidence: Data enters via
f.searchMarketsandf.getMarketinindex.ts; no boundary markers are present; capabilities are limited to console output; sanitization includes basic string slicing.
Audit Metadata