The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill guides the agent to ingest and analyze untrusted data from project files such as package.json and various configuration files.
Ingestion points: package.json, tsconfig.json, .eslintrc*, and lock files.
Boundary markers: No specific delimiters or safety instructions are provided for parsing these files.
Capability inventory: The skill does not perform actions itself but suggests the agent use standard development tools like npm, yarn, or pnpm based on file detection.
Sanitization: No sanitization logic is included for the data read from these files.
[Data Exposure] (SAFE): The skill identifies .env files as common configuration files. This is standard behavior for web development assistants and does not constitute a finding as no exfiltration or unauthorized access patterns are present.
[Command Execution] (SAFE): Mentions of commands like npm install or npm test are instructional and appropriate for a web development context.

web-projects