easy-cheese-setup

Warn

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a Python script via python3 <skill>/scripts/easy-cheese-setup.pyz with commands such as global, local, and doctor. It also invokes hallouminate init-repo to register repository tenants.
  • [DYNAMIC_EXECUTION]: The primary logic is distributed as a Python Zip Application (.pyz) at scripts/easy-cheese-setup.pyz. This is a compressed archive format that bundles source code and dependencies into a single file, making the logic opaque to simple inspection during skill review.
  • [INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection as it processes untrusted data from the environment and file system.
  • Ingestion points: Reads the Hallouminate configuration file (defaulting to ~/.config/hallouminate/config.toml) and scans repositories for .cheese/ artifacts.
  • Boundary markers: The skill uses # >>> easy-cheese:cheese-durable and # <<< easy-cheese:cheese-durable as markers to delimit the block it manages within the configuration file.
  • Capability inventory: The skill has the capability to rewrite configuration files, create directories on the file system, and execute repository initialization commands.
  • Sanitization: The instructions describe the use of markers to prevent overwriting the entire file, but they do not explicitly detail sanitization of the content read from the configuration file before it is processed or written back.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 10, 2026, 11:29 PM
Security Audit — agent-trust-hub — easy-cheese-setup