Orchestrate agents · Automate reviews · Streamline Git · Ship with confidence
gtools is a collection of seven skills for Claude Code. It brings Git and pull-request automation, iterative code reviews, multi-agent task orchestration, and timesheet management into your development workflow.
Designed to reduce repetitive coordination while keeping review gates and explicit approvals where they matter.
| Skill | What it does |
|---|---|
| team-lead | Coordinate two developers and a reviewer with task dependencies, worktrees, and linear merges. |
| pre-commit-review | Run a review-and-fix loop before allowing a commit. |
| post-push-review | Watch CI review results, fix reported issues, and push again. |
| squash-branch | Consolidate branch commits into one after showing a plan and obtaining approval. |
| list-unresolved-comments | List unresolved review threads on the current pull request. |
| resolve-comments | Resolve unresolved or outdated pull-request review threads in bulk. |
| update_timesheets | Calculate and submit BambooHR timesheet entries. |
Note
The pre-commit and post-push hooks run in an active Claude Code workflow. They are not a replacement for your repository's CI policies or human approval requirements.
- Claude Code (CLI or desktop)
- Git
- GitHub CLI (
gh) for GitHub-related skills
Option A — Direct installation (no clone required)
claude --plugin-url https://github.com/argami/gtools/archive/refs/heads/main.zipOption B — Personal plugin marketplace
Run these commands inside Claude Code:
/plugin marketplace add argami/gtools
/plugin install gtools
Option C — Community marketplace (subject to approval)
/plugin install gtools@claude-community
Option D — Local development
claude --plugin-dir /path/to/gtoolsVerify: Type /gtools: in Claude Code. Autocomplete should display the seven available skills.
| Feature | Requirement |
|---|---|
| GitHub and PR skills | Authenticated gh CLI with appropriate repository permissions |
| Team orchestration | Claude Code Agent Teams enabled |
| Timesheets | BambooHR credentials saved to ~/.timesheets_config.json after initial setup |
| TypeSafe integration | Optional TYPESAFE_API_KEY environment variable |
| Review hooks | Plugin installed and hooks enabled |
To enable Agent Teams for team-lead:
export CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1
claude --teammate-mode autoSelected skills integrate with TypeSafe for calibrated semantic decisions rather than relying solely on keyword matching or inferred intent.
export TYPESAFE_API_KEY=your_key_here| Skill | TypeSafe responsibility | Fallback without API key |
|---|---|---|
update_timesheets |
Detect dry-run intent across phrasings and languages | Keyword matching |
squash-branch |
Classify commit-message conventions | Infer from Git history |
squash-branch |
Interpret user approval or rejection | Yes/no heuristic |
team-lead |
Parse reviewer verdicts | Match APPROVED / CHANGES REQUESTED |
TypeSafe is optional: all supported skills retain their fallback behavior.
Contributions and new skills are welcome.
- Create
skills/<skill-name>/. - Add a
SKILL.mdwith frontmatter metadata and step-by-step instructions. - Add a detailed
README.mdfor complex skills. - Update the version in
.claude-plugin/plugin.jsonwhen publishing. - Test the skill via
/gtools:<skill-name>in Claude Code.
See CLAUDE.md for development conventions and skill structure.
Found an issue or have an idea? Open a GitHub issue.
Released under the MIT License.