write-a-skill
Pass
Audited by Gen Agent Trust Hub on May 19, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill establishes a process for generating new AI agent skills and utility scripts based on untrusted user input. This creates a surface for indirect prompt injection (Category 8).\n
- Ingestion points: User requirements are gathered in the 'Gather requirements' phase (SKILL.md).\n
- Boundary markers: The provided SKILL.md template lacks explicit delimiters or instructions to treat user-provided data as untrusted, which could lead to instructions being executed as logic in the generated skill.\n
- Capability inventory: The process explicitly allows for the creation of utility scripts (e.g.,
scripts/helper.js) which could contain malicious logic if derived from malicious requirements.\n - Sanitization: There are no instructions for validating, escaping, or sanitizing the user input before it is used to draft the skill components.
Audit Metadata