patsnap-lifescience-pharmaceuticals-exploration
Pass
Audited by Gen Agent Trust Hub on Apr 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its core function of ingesting and summarizing untrusted data from external sources such as news, academic papers, and general web searches. Such data could contain adversarial instructions intended to influence the agent's behavior during report generation.
- Ingestion points: Untrusted content is retrieved through tools like
ls_news_fetch,ls_paper_fetch, and the general web search tool (SKILL.md). - Boundary markers: The instructions lack explicit delimiters or warnings to the agent to disregard any instructions found within the retrieved data (SKILL.md).
- Capability inventory: The skill utilizes a suite of data retrieval tools (
ls_*) for patents, clinical trials, and drug data, and uses these to generate synthesized reports. It does not possess dangerous execution capabilities like shell access or file writing. - Sanitization: There is no mention of sanitization, filtering, or validation of the fetched data before it is integrated into the prompt context.
Audit Metadata