The Agent Skills Directory

Project Command Execution: The skill instructs the agent to run existing project scripts (build, lint, test) and TypeScript compiler commands to verify type-safe repairs. This is an expected behavior for development tasks.
Secure Dependency Handling: The instructions explicitly direct the use of frozen or immutable installation commands (e.g., npm ci, yarn install --immutable). This practice maintains project integrity by preventing accidental lockfile mutations.
Mitigation of Package Confusion: The skill includes a specific warning against using npx tsc for availability tests, which helps avoid common ecosystem risks like package name confusion.
Indirect Prompt Injection Surface: The skill processes project configuration and source files. Ingestion points include package.json, tsconfig.json, and project source code. Capability inventory includes file system writes and shell command execution (npm, yarn, pnpm, tsc). Boundary markers and sanitization are not explicitly defined, as the skill's logic is constrained to technical type-system analysis and standard development workflows.

typescript-strict