git-secure

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install git-crypt and age using the Homebrew package manager (brew install). These are established open-source tools from a well-known service.
  • [COMMAND_EXECUTION]: The skill executes several shell commands to manage repository state and encryption, including git-crypt init, git-crypt status, git add, git tag, and tar. It also suggests the use of git filter-repo for history scrubbing, which is a destructive operation that the skill correctly identifies with a mandatory user confirmation warning.
  • [PROMPT_INJECTION]: Indirect surface exists as the skill ingests user-provided folder paths and glob patterns which are then interpolated into shell commands for archival (tar) and status checks (git-crypt status).
  • Ingestion points: User-provided paths and patterns in the "What to encrypt" step.
  • Boundary markers: Not explicitly defined for command interpolation.
  • Capability inventory: Subprocess execution for git, brew, tar, and age (SKILL.md).
  • Sanitization: No explicit sanitization or validation of user-provided paths is mentioned before execution.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 03:28 AM