playbook-fundraising
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill architecture is secure, isolating data processing in JavaScript and Python nodes before the model receives data.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests company data from external search APIs. 1. Ingestion points: External data enters through {{Funding Raw}} in SKILL.md and clay-table.md. 2. Boundary markers: The prompt in SKILL.md enforces a strict JSON schema and explicit constraints on content creation. 3. Capability inventory: Code nodes in clay-workflow.md (Node 3, 5) handle logic; the LLM (Node 4) has no file-write or network-access capabilities. 4. Sanitization: Extensive filtering in clay-table.md (Column 2) and clay-workflow.md (Node 3) validates domains, funding stages, and dates.
- [COMMAND_EXECUTION]: Documentation provides standard CLI examples for platform setup; no malicious or autonomous execution detected.
- [DATA_EXPOSURE]: The skill accesses public lead data via platform lookups; no private credential harvesting or sensitive file access is present.
Audit Metadata