modern-git

Pass

Audited by Gen Agent Trust Hub on Oct 3, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONPRIVILEGE_ESCALATIONPERSISTENCEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill contains instructions to download and execute shell scripts from gitbutler.com and claude.ai using the curl | sh and curl | bash patterns.\n- [PRIVILEGE_ESCALATION]: The documentation includes commands using sudo to move the but CLI binary to /usr/local/bin for system-wide installation.\n- [PERSISTENCE]: The skill provides instructions to modify user shell profiles, such as ~/.bashrc and ~/.zshrc, to persistently update the PATH environment variable with the location of local binaries.\n- [INDIRECT_PROMPT_INJECTION]: The skill describes a feature where branch context and history are automatically injected into AI agent prompts during collaborative sessions, creating a potential surface for indirect injection from untrusted repository content.\n
  • Ingestion points: Branch context and commit history are pulled into the agent's prompt during AI sessions.\n
  • Boundary markers: None identified in the provided instructions.\n
  • Capability inventory: The skill facilitates general Git operations and interactive AI sessions.\n
  • Sanitization: Not specified in the documentation.\n- [EXTERNAL_DOWNLOADS]: The skill references external downloads for installation scripts and binaries from the gitbutler.com and claude.ai domains.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 3, 2026, 10:28 AM