create-skill
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is purely instructional and provides templates for scaffolding new skills. It does not perform any unauthorized network requests, credential harvesting, or privileged operations. All referenced files are local and match the provided source code.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided descriptions and parameters to generate new skill files. This represents a standard input surface for an AI agent and is necessary for its primary function.
- Ingestion points: User descriptions and arguments provided during the skill generation process in SKILL.md.
- Boundary markers: Not explicitly included in the generated output templates, though the instructions guide the AI to use structured markdown.
- Capability inventory: The skill facilitates file system operations via the agent's core tools (file reading and writing) to create the skill structure.
- Sanitization: Relies on the underlying LLM's interpretation of instructions rather than explicit programmatic filtering.
Audit Metadata