cold-outreach
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and agent manifests define a structured and benign workflow for content creation. The multi-layer refinement process, which includes a Critic agent and a Voice Auditor, provides strong guardrails against common AI writing failures and ensures the output remains professional.
- [DATA_EXPOSURE]: The skill is configured to read project-specific files such as product-context.md and icp-research.md. This data exposure is necessary for the skill to perform personalization and is performed within the scope of the agent's research tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it retrieves and processes untrusted data from external sources like LinkedIn profiles, Twitter/X posts, and web search results. However, the risk is mitigated by secondary agents that review the generated content before it is presented to the user.
- [COMMAND_EXECUTION]: The skill manifest allows the use of the Bash tool. The analysis of the instructions and scripts did not reveal any commands that attempt to perform privilege escalation, establish persistence, or execute malicious payloads.
Audit Metadata