doc-coauthoring
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external, potentially untrusted sources, creating a surface for indirect prompt injection.
- Ingestion points: The workflow encourages users to "link to shared documents," "point to team channels or threads," and use integrations (Slack, Teams, Google Drive, SharePoint) to pull context directly into the agent's environment.
- Boundary markers: The instructions do not specify the use of delimiters or explicit "ignore embedded instructions" warnings when the agent reads content from these external sources.
- Capability inventory: The skill has the capability to write files (
create_file), perform surgical edits (str_replace), and invoke sub-agents to process and test the gathered context. - Sanitization: No explicit sanitization or filtering is mentioned for the content retrieved from external integrations or files, meaning instructions embedded in those documents could potentially influence the drafting or testing phases.
Audit Metadata