The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the nvm (Node Version Manager) installation script from the official nvm-sh GitHub repository and executes it via bash. This is a standard procedure for configuring Node.js environments and uses a well-known community resource.
[COMMAND_EXECUTION]: The skill performs comprehensive dependency management and testing by running npm install, yarn install, or pnpm install and ng test within the target project. While this involves executing code from the audited project (such as lifecycle scripts), it is a necessary functional requirement for generating the health report and test coverage metrics.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data (configuration files, source code, and documentation) to generate its findings. This ingestion of external data into an environment with shell and file-writing capabilities constitutes a potential attack surface. However, the risk is managed by the skill's modular architecture and specific focus on technical metadata.
Ingestion points: Reading package.json, angular.json, tsconfig.json, .eslintrc, and all source files in the project.
Boundary markers: None explicitly defined for untrusted project data.
Capability inventory: Uses Bash, Agent, Task, Write, Edit, and Grep across multiple analysis waves.
Sanitization: No specific sanitization of project-derived content is described before processing by AI agents.

angular-health-audit