The Agent Skills Directory

[SAFE]: The skill facilitates the configuration of the agent.pw security library. It demonstrates best practices for credential management, including the use of environment variables for encryption keys and database connection strings, and local data persistence via PGlite.\n- [COMMAND_EXECUTION]: SKILL.md provides a benign shell command using Node.js to generate a 32-byte base64-encoded encryption key required for the library's initialization.\n- [PROMPT_INJECTION]: The skill documents ingestion points for external data during OAuth authentication flows and API response classification, which creates an indirect prompt injection surface.\n
Ingestion points: The callbackUri parameter in connect.completeOAuth and the response object in connect.classifyResponse (references/api-surface.md).\n
Boundary markers: Not explicitly defined in the provided integration code snippets.\n
Capability inventory: File system write operations for database storage (createLocalDb) and network request capabilities for OAuth token exchanges.\n
Sanitization: Relies on the internal validation and parsing logic of the agent.pw library.

setup-agentpw