kmp-token-saver

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes scripts for installing developer tools using standard package managers such as pip, brew, and winget. These operations are used for the legitimate setup of token-saving utilities.
  • [EXTERNAL_DOWNLOADS]: The skill downloads configuration and packages from official registries and verified sources. Specifically, it fetches the Headroom package from PyPI and RTK via Homebrew or WinGet. The installation of the Ponytail plugin is performed through the official Claude Code marketplace.
  • [SAFE]: The skill demonstrates high security awareness by explicitly forbidding the agent from executing unverified remote scripts (e.g., the Caveman installer) and instructing the user to perform sensitive configuration changes—such as API key management and global hook wiring—manually. It also implements a two-phase authorization model for tool setup, ensuring that persistent environment changes are previewed by the user before being applied.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 03:04 PM