The Agent Skills Directory

[SAFE]: The skill instructs the agent to reference official and well-known sources such as the Angular blog, official documentation, and the main Angular GitHub repository for all API claims, which are recognized as trusted platforms.
[SAFE]: The instructions include specific safety directives that mitigate common risks, such as explicitly forbidding the invention of APIs, requiring the validation of changes using only existing project scripts, and discouraging broad refactors in favor of targeted changes.
[PROMPT_INJECTION]: The skill processes project-specific data which could serve as a surface for indirect prompt injection.
Ingestion points: The skill ingests package.json, lockfile, angular.json, tsconfig, devserver logs, and MCP/WebMCP tool definitions from the local environment.
Boundary markers: Present. The skill mandates using only repository-local validation commands and strictly forbids inventing new APIs or dependency requirements.
Capability inventory: The agent is capable of executing build, test, and lint scripts defined within the user's repository.
Sanitization: The procedure requires manual inspection of inputs to confirm risks before acting, though it lacks automated sanitization steps for the ingested file content.

Angular v22 AI Tutor Safe Usage