account-intelligence-analyst
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
clayCLI tool to run environment checks (clay whoami) and retrieve the available actions catalog (clay workflows actions list). These are legitimate uses of vendor-provided infrastructure for platform integration. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources including public websites, job postings, and news articles. * Ingestion points: Catalog data from /tmp/catalog.json, site content via scraper arms, and external signals via research arms (e.g., job boards, news aggregators). * Boundary markers: The skill uses a structured grading framework (supports/contradicts/not_observed) and requires evidence linking, but does not specify explicit text delimiters for external content. * Capability inventory: Execution of
clayCLI commands and writing to the /tmp directory. * Sanitization: No explicit sanitization or escaping of external content is mentioned. The risk is mitigated by the weighted proxy logic that prevents single-source influence.
Audit Metadata