By mjenkinsx9
Auto stage → secrets-scan → commit → push after every agent turn, strictly per-repo opt-in (/autogit on). One-command undo, quiet batching, PR mode.
When your agent finishes a turn, autogit stages, scans, commits, and pushes. Automatically. Claude Code, Codex, Cursor, and Pi.
autogit wires itself into your agents' lifecycle hooks once, per machine. After that, every agent turn in an opted-in repo ends with stage → secrets scan → commit → push — and git log reads like the instructions you gave your agent, because the turn's prompt becomes the commit subject.
Credit: autogit began as davidondrej/autogit by David Ondrej — all credit to him for the idea and the original MVP. This repo is a heavily reworked fork (v0.5.0): per-clone config that can't leak to teammates, fail-closed safety, quiet batching, PR mode, and a full test suite.
| Feature | What it does | |
|---|---|---|
| 🚢 | Auto-ship | Every agent turn ends with stage → scan → commit → push — zero ceremony |
| ✉️ | Prompts as commit messages | The subject is what you asked your agent to do; "yes"-type replies and slash commands never make it in |
| 🔐 | Secrets gate | Pattern scan over the staged diff blocks the push and unstages — fail-closed |
| ↩️ | One-command undo | autogit undo rewinds the remote and the local commit, leaving your changes back in the working tree |
| ⏱️ | Quiet batching | "quiet": "5m" accumulates turns and ships one commit after the repo goes quiet |
| 🔀 | PR mode | "pr": true pushes to autogit/<branch> and auto-opens a pull request via gh |
| 🤝 | Parallel-agent aware | Busy markers make simultaneous agents take turns; worktrees stay fully isolated |
| 🛡️ | Fail-safe by design | Hooks never disturb the agent; failed pushes are remembered and retried; merge/rebase states are never touched |
/plugin marketplace add mjenkinsx9/mjenkins-toolbox
/plugin install autogit@mjenkins-toolbox
The plugin wires the hooks automatically — no autogit setup, no edits to ~/.claude/settings.json, and disabling the plugin unwires everything. Then opt in per repo:
/autogit on
(/autogit status, /autogit undo, /autogit off, /autogit flush, and /autogit dry-run are also available.) Requires node on PATH — without it the hooks silently no-op.
git clone https://github.com/mjenkinsx9/autogit && cd autogit && npm link
autogit setup
cd your-project && autogit on
setup wires the lifecycle hooks for every agent it finds — Claude Code, Codex, Cursor, and Pi. Every agent turn in an opted-in repo now ships. Repos without autogit on are never touched.
macOS and Linux. Windows is unsupported — the hook commands are POSIX shell. Full install paths, the double-wiring guard, and per-agent notes are in docs/02-installation.md.
Full guides, configuration, and internals live in docs/.
| Doc | Description |
|---|---|
| Overview | What autogit is, the feature set, and platform support |
| Installation & Quick Start | Install as a plugin or the CLI, opt in per repo, supported agents |
| Commands | Every command, ship flags, commit-message rules, and undo |
| Batching & PR Mode | Quiet batching (quiet) and PR mode (pr) |
| Configuration | autogit.json keys and defaults |
| Safety | Opt-in model, secrets scan, undo, merge/rebase guard |
| Install Across Harnesses | Per-harness plugin manifests and auto-ship hooks |
| Internals | Design, how ship/undo/batching work, busy markers, fail-safes |
Full documentation map: docs/README.md
MIT — original work © David Ondrej, modifications © Mike Jenkins.
Matches all tools
Hooks run on every tool call, not just specific ones
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub mjenkinsx9/mjenkins-toolbox --plugin autogitPersonal code review: deterministic CLI gathers diff/context/learnings; the active Claude session reasons over them.
Author, evaluate, security-review, and autonomously improve Claude Agent Skills. Bundles the skill-building workflow as the improving-skills autoresearch loop, goal-setting commands, reference protocols, and the static eval harness.
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).
v9.44.1 — Patch release for Gemini environment/version detection and qwen auth gating. Run /octo:setup.
Unity Development Toolkit - Expert agents for scripting/refactoring/optimization, script templates, and Agent Skills for Unity C# development
Claude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.