The Agent Skills Directory

[SAFE]: Analysis found no evidence of malicious behavior such as unauthorized data access, persistence, or privilege escalation. The skill performs standard development tasks, including file system writes and local command execution via pnpm test to validate new lint rules.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests user-supplied rule names ($ARGUMENTS) and content from existing codebase files to generate new source code. Ingestion points: User-provided arguments and local ESLint rule files. Boundary markers: None. Capability inventory: Local file writing and command execution. Sanitization: None identified. This is considered a low-risk surface inherent to code-generation utilities.

lint-new