skills/cosmix/loom/loom-git-workflow/Gen Agent Trust Hub

loom-git-workflow

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: No attempts to override agent behavior or bypass safety guidelines were found. The instructions are strictly limited to Git workflow guidance.
  • [DATA_EXFILTRATION]: No sensitive file paths (e.g., SSH keys, credentials) are accessed. No network operations or external data transfer mechanisms are present in the command examples.
  • [REMOTE_CODE_EXECUTION]: The skill does not download or execute remote scripts. It provides standard local Git commands for version control management.
  • [COMMAND_EXECUTION]: While the skill lists Bash in its allowed-tools and provides many shell command examples, these are standard Git operations (e.g., git commit, git merge, git worktree) used for their intended purpose in development workflows.
  • [CREDENTIALS_UNSAFE]: No hardcoded credentials, API keys, or secrets are present. Examples use placeholder values and standard Git configuration patterns.
  • [OBFUSCATION]: No obfuscated content, such as Base64 encoding, zero-width characters, or homoglyphs, was detected in the skill file.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 07:34 AM
Security Audit — agent-trust-hub — loom-git-workflow