The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes patterns for an agent to implement IPC handlers that ingest and process external data, creating a potential attack surface if not properly sanitized.
Ingestion points: The open-external IPC channel in emain/emain-ipc.ts accepts a url string from the frontend process.
Boundary markers: The skill does not provide instructions for using boundary markers or explicit instructions to ignore embedded commands within the processed data.
Capability inventory: The skill implements high-privilege capabilities including electron.shell.openExternal(url) for launching system processes and webContents.capturePage(rect) for screen data extraction.
Sanitization: The provided code templates do not include URL validation or sanitization logic before execution, which could allow for the execution of malicious URI schemes (e.g., file:// or javascript:) if untrusted input is passed.
[COMMAND_EXECUTION]: The skill provides instructions for creating IPC handlers that execute system-level operations, such as capturing screenshots (capture-screenshot) and interacting with the OS shell (open-external).

electron-api