buyer-classification
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute
clay whoamito verify the presence and configuration of the necessary vendor-provided CLI tool. This is a benign environment check. - [INDIRECT_PROMPT_INJECTION]: The skill processes contact titles from external lists, creating a potential surface for indirect prompt injection if a title contains instructions intended to manipulate the agent's judgment.
- Ingestion points: Contact job titles are ingested for classification as described in SKILL.md Step 2.
- Boundary markers: The skill enforces a requirement for a rationale that quotes the title evidence, which helps to ground the model's reasoning in the data.
- Capability inventory: The skill does not have network access or capabilities for sensitive file modification; it is designed for data classification and reporting.
- Sanitization: Job titles are normalized (case and punctuation) prior to classification to reduce noise.
Audit Metadata