skills/raine/workmux/worktree/Gen Agent Trust Hub

worktree

Pass

Audited by Gen Agent Trust Hub on Apr 19, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to execute workmux commands and standard shell utilities like mktemp and cat. This behavior is consistent with its stated purpose of managing Git worktrees.
  • [DATA_EXFILTRATION]: The skill explicitly instructs the agent to avoid reading, searching, or exploring the codebase ('HARD RULE — NO EXCEPTIONS: Do NOT explore, read, grep, glob, or search'). This significantly reduces the risk of accidental or malicious data exposure.
  • [INDIRECT_PROMPT_INJECTION]: While the skill processes user-supplied task descriptions to generate prompts for other agents, it uses a secure shell pattern (cat > "$tmpfile" << 'EOF') in its implementation logic. Quoting the heredoc delimiter ('EOF') ensures that user content is treated as a literal string and prevents shell expansion or command injection during the file-writing process.
  • [REMOTE_CODE_EXECUTION]: No evidence of external downloads, unverified package installations, or remote script execution was found. All operations are performed locally using the provided workmux tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 19, 2026, 08:35 AM