doc-coauthoring
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data from user-provided files, templates, and shared documents through messaging and storage integrations during the 'Context Gathering' stage. Because these external inputs are processed without strict boundary markers or instructions to ignore embedded commands, there is a vulnerability to indirect prompt injection where instructions hidden in those documents could influence the agent's behavior.
- Ingestion points: Identified in
SKILL.md(Stage 1) where the agent reads files, templates, and shared documents. - Boundary markers: Absent; the instructions do not define specific delimiters or warnings to ignore instructions inside gathered context.
- Capability inventory: The skill utilizes file modification tools such as
create_fileandstr_replace, and invokes sub-agents for testing (SKILL.md). - Sanitization: Absent; the skill does not specify any validation or filtering for the ingested data.
Audit Metadata