alterlab-fda
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill fetches data from the external openFDA API, which is then returned to the agent's context. This creates a potential surface for indirect prompt injection if a record in the database were to contain malicious instructions designed to be followed by an AI agent.
- Ingestion points: Data is retrieved in
scripts/fda_query.pyusingrequests.getfrom variousapi.fda.govendpoints. - Boundary markers: The skill does not implement specific delimiters or instructions to the agent to treat the fetched JSON as non-executable data.
- Capability inventory: The skill allows access to
Bash(python:*)andBash(curl:*), which increases the potential impact if a prompt injection were to occur. - Sanitization: No sanitization or filtering of the response content is performed before the data is passed to the agent.
- [EXTERNAL_DOWNLOADS]: The skill performs network requests to
api.fda.gov. This is a well-known service provided by the US Food and Drug Administration and is required for the skill's primary functionality.
Audit Metadata