figure-it-out
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ground and process high-level user tasks such as 'large migrations' or 'ambitious multi-part changes' to generate workflow phases.
- Ingestion points: User-supplied task descriptions are interpreted in SKILL.md (Phase A and B).
- Boundary markers: The skill lacks explicit delimiters or instructions for the agent to ignore embedded commands within user-provided project descriptions.
- Capability inventory: The skill references multiple other internal skills and encourages the use of repository scripts to generate audit evidence.
- Sanitization: No explicit sanitization of user input is mentioned before it is used to structure the agent's workflow.
- [COMMAND_EXECUTION]: Phase D encourages the execution of 'committed scripts' within the repository to produce evidence for the audit trail. This behavior relies on the security and integrity of scripts already present in the user's environment.
- [NO_CODE]: The skill consists exclusively of instructional markdown without accompanying executable script files (e.g., .py or .js).
Audit Metadata