pubmed-search
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using
findandheadto resolve the absolute path of its internal scripts within the environment's plugin cache directory. - [EXTERNAL_DOWNLOADS]: Documentation references external vendor libraries
valyu(PyPI) andvalyu-js(NPM) for application integration. - [DATA_EXFILTRATION]: The skill transmits user-provided search queries and an API key to the vendor's official API endpoint at
api.valyu.aito retrieve search results. - [PROMPT_INJECTION]: The skill ingests and processes third-party content from the PubMed database. This represents an indirect prompt injection surface where external research data could potentially contain instructions aimed at influencing the agent's behavior.
- Ingestion points: External search results retrieved via the
scripts/searchtool. - Boundary markers: None explicitly defined in the provided documentation for search result output.
- Capability inventory: The skill utilizes
scripts/search(bash/node) and the ability to output results to the agent's context. - Sanitization: Not specified in the provided instructions.
Audit Metadata