subject-line-lab
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external input including campaign CSV exports, CRM tokens, and competitor subject lines. This ingestion of arbitrary text creates an attack surface where malicious instructions could be embedded to influence the agent's behavior.
- Ingestion points: Campaign CSV exports, pasted subject lists, competitor content, and CRM personalization tokens.
- Boundary markers: The instructions explicitly direct the agent to treat these inputs as untrusted and ignore any embedded commands, referencing a global security policy.
- Capability inventory: The skill performs file writes to the 'memory/' directory and interacts with the 'registry-events.py' script.
- Sanitization: The skill relies on prompt-based instructions to ignore malicious content rather than technical sanitization or schema validation.
- [DYNAMIC_EXECUTION]: The skill instructions specify a workflow involving the execution of a local script 'registry-events.py' to record offer claims.
- Execution Method: Interacts with local scripts to perform stateful updates to the registry.
- Controls: The skill mandates explicit human authorization before the script is used to perform a write operation, serving as a critical safety check.
Audit Metadata