The Agent Skills Directory

[SAFE]: The skill provides comprehensive guidelines for TypeScript development, including discriminated unions, branded types, and constructive modeling. These practices improve code robustness and minimize runtime errors.
[SAFE]: The instructions emphasize using unknown over any for external data and performing boundary validation (parseUser). These are recommended security practices for handling untrusted input.
[INDIRECT_PROMPT_INJECTION]: While the skill instructions apply to the processing of source code files (.ts, .tsx), the guidelines explicitly encourage the agent to implement rigorous validation and narrowing of untrusted data, effectively mitigating risks associated with malicious payloads in data streams.
Ingestion points: Processing of .ts and .tsx files during read/edit operations.
Boundary markers: The skill refers to a separate boundary-discipline skill for managing data entry points.
Capability inventory: File system access for reading and writing source code.
Sanitization: Promotes the use of unknown types and explicit parse functions at data boundaries.

typescript-best-practices