doc-coauthoring
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection due to its core functionality of ingesting and acting upon untrusted external data.
- Ingestion points: As described in
SKILL.md, the workflow retrieves content from external integrations (Slack, Teams, Google Drive, SharePoint), user-provided links, and info dumps. - Boundary markers: The instructions lack defined delimiters or specific prompts to ignore instructions that might be embedded within the ingested external data.
- Capability inventory: The skill has access to tools for creating and editing files (
create_file,str_replace) and can spawn sub-agents for reader testing, which could be influenced by injected instructions. - Sanitization: There is no evidence of content sanitization or validation performed on data retrieved from external sources before it is processed.
Audit Metadata