The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard build and type-checking utilities, specifically tsc, nx, and bun, to perform its primary function of gathering type error reports.
[PROMPT_INJECTION]: The skill processes untrusted content from the codebase, creating an indirect prompt injection surface. However, this risk is mitigated by strict reasoning protocols and human oversight.
Ingestion points: Processes project files and codebase search results.
Boundary markers: Relies on internal protocols (e.g., evidence-based reasoning) rather than explicit delimiters.
Capability inventory: Ability to execute build commands and modify local files.
Sanitization: Changes require explicit user confirmation via AskUserQuestion and are gated by confidence scores and file-line evidence.
[SAFE]: All operations are consistent with the skill's stated purpose, focusing on internal project maintenance without external data transmission or unauthorized system access.

fix-types