swap-test
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from emails and lead data to perform sentence-level analysis and rewriting (Step 1, Step 3).
- Ingestion points: Untrusted data enters the agent context through 'The email' and 'Lead data' (SKILL.md).
- Boundary markers: Absent; there are no instructions to ignore commands embedded within the emails or data rows.
- Capability inventory: The agent can execute shell commands via the 'clay' CLI tool and write files to the working directory (Step 0, Step 7).
- Sanitization: Absent; no validation or escaping of the input content is described before processing.
- [COMMAND_EXECUTION]: The skill utilizes the 'clay' CLI to verify the execution environment and retrieve table data (Step 0). It specifically instructs the agent to run
clay --version,clay whoami, andclay tables list, which are legitimate operations given the skill's purpose as a Clay-integrated tool.
Audit Metadata