doc-coauthoring
Pass
Audited by Gen Agent Trust Hub on Apr 18, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill uses instructional language to define a structured collaborator persona, which is appropriate for its documented purpose and does not attempt to bypass agent safety filters or override core instructions.- [INDIRECT_PROMPT_INJECTION]: A potential surface for indirect injection exists within the 'Reader Testing' stage, where user-provided or externally fetched content is processed by a sub-agent.
- Ingestion points: The skill ingests text from user info-dumps and shared documents via platform integrations (e.g., Google Drive, Slack).
- Boundary markers: The instructions do not explicitly mandate the use of XML delimiters or 'ignore instructions' warnings when passing content to the testing sub-agent.
- Capability inventory: The skill maintains the ability to create and modify files and invoke sub-agents for analysis.
- Sanitization: The workflow relies on the inherent guardrails of the sub-agent and user oversight rather than explicit data sanitization.- [DATA_EXPOSURE]: The skill references the use of official platform connectors and integrations to fetch project context. These operations are performed with user consent and through established security boundaries, representing standard functional behavior for a documentation assistant.
Audit Metadata