clean-git-repo
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill orchestrates the execution of a bundled Bash script (
scripts/clean-git-repo.sh) to perform repository maintenance. The script utilizes standard development tools includinggitand the GitHub CLI (gh) to query branch status and perform deletions. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from the local file system and remote APIs which could theoretically contain malicious instructions designed to influence the agent.
- Ingestion points: Branch names, worktree paths, and GitHub Pull Request metadata are ingested via
git for-each-ref,git worktree list, andgh pr listin thescripts/clean-git-repo.shfile. - Boundary markers: The skill's instructions require the agent to generate a dry-run report and obtain explicit user confirmation before applying deletions, acting as a manual safety boundary.
- Capability inventory: The skill is capable of deleting local branches and worktrees, pruning remote tracking refs, and performing remote branch deletions via
git push. - Sanitization: The underlying Bash script implements secure coding practices, including comprehensive variable quoting and the use of the
--delimiter for Git commands, which prevents command injection from maliciously named branches or paths.
Audit Metadata