narrated-walkthrough-to-a-numbered-sop-document
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration attempts were detected. The skill's behavior is consistent with its stated purpose of document generation and task logging.
- [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing dependencies using
npxandopenclawfrom the vendor's official GitHub repository (AgentPMT/agent-skills). These are standard installation methods for the AgentPMT ecosystem. - [PROMPT_INJECTION]: The workflow processes untrusted transcript data from the Plaud tool, which creates an indirect prompt injection surface.
- Ingestion points: Fetching transcripts via the
plaudtool integration documented inSKILL.md. - Boundary markers: None explicitly defined to separate instructions from transcript content.
- Capability inventory: The agent has permissions to write to Google Docs and Google Sheets through integrated connectors.
- Sanitization: No explicit validation or filtering of transcript content is performed before processing.
- This represents a standard architectural risk for data-transformation skills, mitigated by the specific scope of the output tools.
Audit Metadata