skill-tutor
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run
node .skill-sandbox/validations/validate-sandbox-skills.mjs. This execution is confined to the local sandbox and is intended for validating the user's progress in the tutorial. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-supplied input through structured worksheets, creating a potential surface for indirect instructions. 1. Ingestion points: Input fields for 'learning goal', 'preferred skill type', and 'available time', as well as various free-text worksheet fields in SKILL.md. 2. Boundary markers: No clear delimiters are provided to separate user-supplied data from the instruction context. 3. Capability inventory: The skill has the ability to create directories (
mkdir), change directories (cd), and execute local Node.js scripts (node). 4. Sanitization: There is no evidence of filtering or escaping user-provided text before it is utilized in the guided process.
Audit Metadata