drug-discovery-search
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from external databases such as ChEMBL, DrugBank, and FDA drug labels. This content is ingested into the agent's context and could potentially contain instructions or deceptive content that targets the agent's behavior.
- Ingestion points: Results retrieved from the Valyu API and processed by the
scripts/searchscript. - Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined to isolate external data.
- Capability inventory: The skill executes shell scripts and performs file system searches using the
findcommand. - Sanitization: There is no mention of sanitization, filtering, or validation of the external content before it is returned to the agent.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to locate and run its search utility.
- Evidence: The command
find ~/.claude/plugins/cache -name "search" ...is used to resolve the script path at runtime. - This behavior is localized to the skill's own installation environment and is a functional requirement for locating the cached scripts.
Audit Metadata