ketchup
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from chat history, subagent reports, and project metadata.
- Ingestion points: The skill collects work history, reports from subagents, and messages from other users in the chat (see SKILL.md under 'Gather what changed').
- Boundary markers: The instructions explicitly enforce a read-only state, directing the agent to 'do not change code, start an app or a job, or post anything' while processing this data.
- Capability inventory: Actions are restricted to information gathering and generating summaries/visuals (Mermaid diagrams).
- Sanitization: The skill relies on clear instructional boundaries to prevent instructions within the processed chat history from being executed.
Audit Metadata