electron
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill documents a 'file:read' IPC handler that accepts a path from the renderer and reads it via 'fs.readFile' in the main process. Without path validation, this pattern establishes a capability for arbitrary file access if the renderer process is compromised.
- [COMMAND_EXECUTION]: An example 'file:write' IPC handler in SKILL.md allows the renderer to specify both a file path and content for writing via 'fs.writeFile', creating a potential risk for overwriting application or system files if user input is not sanitized.
- [INDIRECT_PROMPT_INJECTION]: The skill defines a communication surface where untrusted data from the renderer is processed by sensitive Node.js APIs. Ingestion points: Arguments passed to the 'file:read' and 'file:write' IPC channels in SKILL.md. Boundary markers: None implemented in the handler examples. Capability inventory: Includes reading and writing to the filesystem via the 'fs/promises' module. Sanitization: The provided examples do not demonstrate path validation, directory sandboxing, or content filtering.
Audit Metadata