known-problem-hint-research
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill directs the agent to interact with a wide array of well-known external technical and security resources to gather information. These include academic repositories (ArXiv, IACR), standards organizations (IETF, W3C, NIST), and security databases (NVD, OSV, CVE).
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it is designed to ingest and process unstructured data from external web sources.
- Ingestion points: SKILL.md (instructions to read sources through connectors, fetch tools, or browser automation).
- Boundary markers: The skill does not implement specific text-based delimiters but advises the agent to load a hygiene skill when processing external input.
- Capability inventory: The skill utilizes web search tools, browser automation, and data extraction capabilities.
- Sanitization: The instructions include explicit safeguards, mandating the removal of secrets from external queries and the inspection of external code for malicious intent before use.
Audit Metadata