skills/darkmice/talon-docs/talon/Gen Agent Trust Hub

talon

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as documentation for the Talon database engine. No prompt injection, data exfiltration, or persistence mechanisms were detected.
  • [EXTERNAL_DOWNLOADS]: The skill references official SDKs hosted on the author's GitHub repository (github.com/darkmice/talon-sdk) and a pre-compiled commercial library from the vendor's domain (releases.talon.dev). These are legitimate vendor resources and do not pose a security risk.
  • [COMMAND_EXECUTION]: Code snippets include standard package manager commands (e.g., go get, npm install, git clone) used to set up the database environment. These are expected for a development-focused skill.
  • [DATA_EXFILTRATION]: All data operations described are local to the user's defined data directory (e.g., ./data). There are no unauthorized network operations or transfers to untrusted third parties.
  • [REMOTE_CODE_EXECUTION]: While the skill mentions loading a pre-compiled commercial library (talon-ai), this is a documented feature of the product provided by the vendor. The SDKs use standard Foreign Function Interface (FFI) mechanisms (like Python's ctypes or Node.js ffi-napi) to interact with the native library, which is routine for high-performance database drivers.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 11:25 PM