using-superpowers
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Uses high-pressure directives in SKILL.md such as 'ABSOLUTELY MUST', 'NOT NEGOTIABLE', and 'YOU DO NOT HAVE A CHOICE' to suppress the agent's autonomy and force specific behaviors regardless of context.
- [INDIRECT_PROMPT_INJECTION]: Mandates a workflow in SKILL.md that forces the ingestion of external data (other skills) with a minimal relevance threshold ('even a 1% chance'), significantly increasing the attack surface for malicious third-party content. Evidence Chain: 1. Ingestion points: Skills are loaded from the environment/catalog as described in SKILL.md using the 'Skill' tool. 2. Boundary markers: Absent; no explicit delimiters or instructions to ignore embedded commands are present in the defined workflow. 3. Capability inventory: The 'Skill' tool is used to load and execute content from external files as specified in SKILL.md. 4. Sanitization: Absent; no review or sanitization is required before the agent is commanded to 'follow skill exactly'.
Audit Metadata