step-back
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill utilizes untrusted conversation history to generate analytical reports.
- Ingestion points: It reads the full dialogue history as part of the restating goals and mapping paths steps in SKILL.md.
- Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the ingested history.
- Capability inventory: The skill has the capability to write the generated analysis to local markdown files (e.g., step-back-YYYY-MM-DD-HHMM.md).
- Sanitization: There is no evidence of data sanitization or validation of the conversation history before it is written to disk.
Audit Metadata