PANews
Pass
Audited by Gen Agent Trust Hub on May 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a local Node.js script (scripts/cli.mjs) as a unified entrypoint for all data retrieval operations, including searching articles and fetching leaderboards.
- [DATA_EXPOSURE]: No evidence of sensitive file access or hardcoded credentials was found. The skill is designed to interact with publicly available news and market data from PANews and Polymarket.
- [PROMPT_INJECTION]: The skill processes untrusted external data, which creates an indirect prompt injection surface. 1. Ingestion points: Full article content via the get-article command and community comments via the get-topic command. 2. Boundary markers: The workflows for article synthesis and topic research do not define explicit delimiters or warnings to isolate external data from agent instructions. 3. Capability inventory: The skill is limited to executing its internal reader CLI for network-based data retrieval; it does not have access to file-system write or arbitrary shell execution capabilities. 4. Sanitization: No specific sanitization or filtering logic for ingested article or comment text is described in the provided workflow documentation.
Audit Metadata