clinical-trials-search
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a Bash wrapper (
scripts/search) to execute a Node.js CLI tool. It also includes instructions for the agent to use thefindcommand to locate its own scripts within the platform's local plugin cache (~/.claude/plugins/cache). - [INDIRECT_PROMPT_INJECTION]: The skill processes clinical trial data fetched from an external API (Valyu), which serves as a potential vector for indirect prompt injection if the source content contains malicious instructions.
- Ingestion points: Data is ingested through search result fields such as
title,url, andcontentreturned by the API. - Boundary markers: The instructions do not specify any delimiters or safety warnings to wrap the interpolated content.
- Capability inventory: The skill has the capability to execute shell commands via the
scripts/searchwrapper. - Sanitization: There is no mention of sanitization, filtering, or validation of the clinical trial data before it is presented to the agent or processed.
Audit Metadata