recipe-update-doc
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources which is then used to construct prompts for other agents.
- Ingestion points: User input via
$ARGUMENTSand document content fromdocs/paths referenced inSKILL.md. - Boundary markers: The skill uses structured prompts for subagents, but instructions to copy content 'verbatim' increase injection risk.
- Capability inventory: It can trigger file updates and verification steps through various subagents as mapped in
SKILL.md. - Sanitization: The process includes mandatory review gates and a final manual approval checkpoint to intercept malicious output.
Audit Metadata