The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): The skill executes 'typecheck', 'lint', and 'test' scripts automatically during the smoke validation phase (SKILL.md Step 10). If a project contains malicious code in its package.json scripts, the agent will execute it on the host system.
[PROMPT_INJECTION] (HIGH): The skill is vulnerable to Indirect Prompt Injection (Category 8) due to its high-privilege capabilities combined with untrusted data ingestion. 1. Ingestion points: Reads 'package.json', lockfiles, and configuration files from the project. 2. Boundary markers: Absent. 3. Capability inventory: Command execution (npm install, npm test) and file system write operations. 4. Sanitization: Absent; the skill merges project-provided scripts with internal templates.
[EXTERNAL_DOWNLOADS] (LOW): The skill installs several third-party Node.js packages (typescript, @biomejs/biome, vitest, husky, lint-staged). While these are industry-standard tools, they are external dependencies fetched at runtime.

ts-starter