evidence-synthesis
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from external literature sources, creating a potential vector for indirect prompt injection if the research papers contain malicious instructions.
- Ingestion points: External academic databases such as OpenAlex, Consensus, PubMed, Scopus, and Cochrane, as well as local resource library folders (
Ane's RESURSE/) (SKILL.md). - Boundary markers: The instructions lack specific delimiters or warnings for the agent to ignore instructions that might be embedded within the research text (SKILL.md).
- Capability inventory: The skill triggers searches through internal literature tools and uses the
Edittool to update files (SKILL.md). - Sanitization: There is no evidence of sanitization or filtering of the external research content before it is synthesized into the final output (SKILL.md).
- [COMMAND_EXECUTION]: The skill instructs the agent to execute specific analysis commands, such as
dup-testandera-split, which are part of a referenced internal literature skill set.
Audit Metadata