create-skill
Pass
Audited by Gen Agent Trust Hub on Jul 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill invokes 'npx -y skills-ref validate' which downloads and executes the 'skills-ref' package from the npm registry. This occurs during the validation step without a pinned version, creating a dependency on an external package source.- [COMMAND_EXECUTION]: The workflow involves executing local Node.js scripts (e.g., 'node scripts/validate-skill-metadata.mjs') and the external 'npx' CLI to perform validation checks on the skill files.- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) as it processes untrusted user data to generate or modify skill files.
- Ingestion points: Workflow Step 1 ingests user-provided task descriptions, inputs, expected outputs, and success criteria in SKILL.md.
- Boundary markers: There are no explicit delimiters or instructions to the agent to disregard instructions potentially embedded within the user-provided use cases.
- Capability inventory: The skill directs the agent to create/modify files across the 'references/', 'assets/', and 'scripts/' directories and execute shell commands for validation.
- Sanitization: No specific sanitization or verification of the user-provided data is performed before it is used to structure the resulting skill.
Audit Metadata