using-superpowers
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes an aggressive policy for discovering and executing external instructions (skills) from the
~/.openclaw/extensions/superpowers/skills/directory with a very low activation threshold ('1% chance'). This pattern increases the attack surface for indirect prompt injection, as the agent is directed to prioritize these external instructions over its default behavior with minimal validation. - Ingestion points: The agent is instructed to search for and read files within the
~/.openclaw/extensions/superpowers/skills/directory at the start of every task. - Capability inventory: The agent possesses the capability to load these files and 'follow instructions exactly,' which can include file operations, command execution, or network access depending on the loaded skill's contents.
- Boundary markers: The skill lacks explicit instructions to sanitize the contents of the loaded skills or to treat them as untrusted data.
- Sanitization: No sanitization or validation logic is defined for the content found in the skills directory.
Audit Metadata