The Agent Skills Directory

[SAFE] (SAFE): The skill is explicitly designed with a security-first approach, providing rules and templates that prevent common Electron vulnerabilities such as context isolation bypasses, hardcoded encryption keys, and insecure IPC communication.\n- [Indirect Prompt Injection] (SAFE): The skill contains a vulnerability surface for indirect injection via custom protocol handlers (deep linking). However, the implementation includes robust mitigations such as strict path routing and mandatory CSRF state parameter validation.\n
Ingestion points: main.ts handles deep links through the app.on('second-instance') and app.on('open-url') events.\n
Boundary markers: The implementation uses cryptographic state validation and explicit path checks (/auth/callback) before processing data.\n
Capability inventory: The skill allows opening external URLs via shell.openExternal, updating local storage via electron-store, and specific main/renderer IPC communication.\n
Sanitization: Incoming protocol strings are parsed using the standard URL constructor to ensure integrity.

electron-base