An IDE for AI CLI Agents

Docs, plans, terminals, autonomous agents, and git — one workspace instead of five scattered tools.

English · 한국어

npm i -g clitrigger && clitrigger

Or download the desktop app — no Node.js needed: Windows .exe · macOS .dmg · Linux .AppImage

Up and running in 60 seconds — open http://localhost:3000, set a password, add a project, write TODOs, hit Start.

---

Docs → Plan → Terminal → Autonomous Tasks → Version Control. One pipeline.

Developing with AI CLI agents (Claude Code, Codex, …) scatters the workflow across disconnected tools: requirements in a note app, plans in another tool, agents across a pile of terminal windows, results in a git client. Editor-centric development has the IDE; CLI-agent-centric development doesn't — so the one ferrying context between tools ends up being you.

CLITrigger is that missing IDE. It connects the whole workflow into a single five-stage pipeline — build project knowledge in Docs, shape it into a plan with the planner & calendar, refine it live in terminal sessions, hand it to multiple AI CLIs (Claude Code · Antigravity · Codex) for parallel autonomous execution in isolated git worktrees, and land the results through the review queue and built-in Git client.

Each stage inherits the context of the one before it — the intent you captured in docs flows all the way to the merge.

flowchart LR
    docs["📚 Docs<br>Project knowledge"] --> plan["🗓 Plan<br>Planner · Calendar"]
    plan --> term["⌨️ Terminal<br>Interactive Sessions"]
    term --> auto["🤖 Autonomous Tasks<br>Parallel Worktrees"]
    auto --> vcs["🔀 Version Control<br>Review Queue · Git"]
    vcs -. lessons feed back into docs .-> docs

Loading

AI CLIs working simultaneously across isolated git worktrees

---

Why CLITrigger?

The tools don't talk to each other. As AI writes more of the code, the developer's real job becomes capturing intent and reviewing output — yet that intent lives in a note app, the plan in another tool, the agents in a stack of terminal windows, and the results in a git client. Editor-centric development solved this decades ago with the IDE. CLI-agent-centric development never got one.

CLITrigger is built as that IDE. Its backbone is a single pipeline — Docs → Plan → Terminal → Autonomous Tasks → Version Control — where each stage is not a separate tool but a consumer of the previous stage's context: docs become plans, plans become the agent's prompt, execution results arrive in a review queue. Intent is never lost between stages.

Inside the pipeline sits the execution machinery:

• Parallel execution — every task runs in its own isolated git worktree, with Claude / Antigravity / Codex working simultaneously
• Scheduling around rate limits — cron-based runs and auto-retry at quota reset make full use of your tokens, even while you're away
• Multi-agent quality — architect / developer / reviewer agents debate before implementation, beating a single agent working alone
• One place to land it — triage every diff in the review queue, then commit, push, and merge in the built-in Git client

---

Features

The features follow the five pipeline stages — 📚 Docs → 🗓 Plan → ⌨️ Terminal → 🤖 Autonomous Tasks → 🔀 Version Control — plus the supporting features underneath. Each feature below has a full guide in the Wiki (↗).

📚 1. Docs — build the knowledge

Docs (File-based Knowledge)

A per-project Obsidian-style knowledge base with a [[wikilink]] graph — inject any file into a prompt, CLI-agnostically. What accumulates here is the input to the whole pipeline. ↗

The Docs tab — browse project markdown with inline preview and a force-directed wikilink graph, then selectively inject files into prompts

🗓 2. Plan — capture the intent

My Schedule

One personal calendar overlaying your memos, every project's schedules, planner due dates, and assigned Jira issues. ↗

One calendar overlaying personal memos, cross-project schedules, planner due dates, and assigned Jira issues

Planner

A lightweight task planner — capture ideas, then convert any item into a TODO, schedule, or session; Markdown import/export. What you plan here becomes the execution unit of the next stage. ↗

Inline editing, color-coded tags, image attachments, and one-click conversion to TODOs or schedules

⌨️ 3. Terminal — refine it with AI

Interactive Sessions

Long-lived CLI sessions in floating windows with VS Code-style docking, pop-out, and real xterm.js terminals — the human-in-the-loop stage before handing work off to automation. ↗

Claude, Antigravity, and Codex sessions docked side-by-side via VS Code-style window grouping — each running in its own worktree branch

🤖 4. Autonomous Tasks — AI executes in parallel

Parallel Worktree Execution (Tasks)

Every TODO runs in its own git worktree with Claude / Antigravity / Codex in parallel, plus dependency chains and merge control. ↗

Multi-Agent Discussion

Architect / developer / reviewer agents debate before implementing, then commit code or send action items to the planner. ↗

Multiple AI agents with different roles debating in the Discussion view

Scheduled Execution

Run tasks on cron or one-off schedules, with auto-retry at the exact rate-limit reset time. ↗

Cron-based recurring and one-time scheduled task execution

Multi-CLI & Sandbox Mode

Pick Claude / Antigravity / Codex per project, TODO, or agent; strict sandbox confines file access to the worktree. ↗

🔀 5. Version Control — review and land it

Morning Review Queue

Triage every overnight TODO across projects in one keyboard-driven card stack — navigate, merge, or discard in a keypress. ↗

Built-in Git Client

A Fork / SourceTree-style Git client in the browser — stage, commit, push, and manage branches and diffs. This is where AI output lands in your history, closing the pipeline. ↗

Commit graph, branch actions, file diffs — all in the browser

🧰 Supporting the pipeline

Analytics

Per-project cost and execution stats — by CLI, by status, and over time. ↗

Cost and token usage broken down by CLI, status, and over time

Live Logs (Chat & Raw)

Real-time WebSocket log streaming in Chat (markdown) or Raw (terminal) mode. ↗

Favorites Launcher

One-click launcher for your frequent external tools (executables, commands, URLs) from the sidebar. ↗

Remote Access

Reach CLITrigger from anywhere via Cloudflare Tunnel, with completion notifications and custom-domain routing. ↗

MCP Server

Expose CLITrigger to any MCP client (Claude Desktop, Claude Code) over an HTTP endpoint — list projects, create and run tasks, and check status just by chatting with your AI. Copy the ready-made config (URL + token) from Settings → MCP; works for npm, desktop, and tunnel users alike. ↗

---

Tech Stack

Layer	Tech
Backend	Node.js · Express · TypeScript · SQLite · WebSocket
Frontend	React 18 · Vite · Tailwind CSS · Recharts
AI CLIs	Claude · Antigravity · Codex (Adapter Pattern)
Git	simple-git (worktree management)
Scheduling	node-cron
Terminal	node-pty (TTY support) · xterm.js (pixel-perfect rendering)
Remote Access	Cloudflare Tunnel (optional)

---

Quick Start

Option A — Desktop App (recommended for end users)

Download the installer for your platform from the latest GitHub release:

• Windows — CLITrigger-Setup-<version>.exe (NSIS installer) or the portable .exe
• macOS — CLITrigger-<version>.dmg (Apple Silicon & Intel)
• Linux — CLITrigger-<version>.AppImage

The desktop app bundles Node.js and the native modules (better-sqlite3, node-pty, cloudflared), so no separate runtime install is needed. On first launch a setup screen appears in the embedded browser — pick a password there and you're in. External sharing (Cloudflare tunnel) stays paused until setup completes, so the first user is guaranteed to be you.

Option B — npm (recommended for developers)

# Install
npm i -g clitrigger
clitrigger

# Upgrade to the latest version
npm i -g clitrigger@latest
# Check current version: clitrigger --version

On first run the server starts immediately. Open http://localhost:3000 → set a password on the welcome screen → register a project → write TODOs → click Start. Change the password later via Settings → Account in the web UI.

CLITrigger also prints a one-line Update available: <new> -> npm i -g clitrigger@latest hint at startup whenever a newer version is on npm — no auto-update, you decide when to upgrade.

# Change settings
clitrigger config port 8080    # Change port
clitrigger config tunnel on    # Enable Cloudflare tunnel for external sharing

Prerequisites: Node.js 22+ (use an LTS release), Git, at least one AI CLI (Claude / Antigravity / Codex)

Supported Platforms: Windows · macOS · Linux — all core code is cross-platform compatible. Prefer an LTS (even-numbered) Node.js. A brand-new major (e.g. an odd/just-released version) may not have prebuilt native binaries yet, which forces a source build requiring a C++ toolchain (Visual Studio Build Tools on Windows, xcode-select --install on macOS).

Run from Source (for development)

Click to expand

# 1. Clone & install
git clone https://github.com/HyperAITeam/CLITrigger.git
cd CLITrigger
npm install
cd src/client && npm install && cd ../..

# 2. Configure environment
cp .env.example .env
# AUTH_PASSWORD is optional — leave it blank and the dev server will show the
# setup screen on first browser load. Set it only if you want to skip setup.

# 3. Run
npm run dev

Open http://localhost:5173.

Windows One-Click Scripts

Double-click any bat file in scripts/ — no terminal needed.

File	Action
install.bat	Install dependencies (first time)
dev.bat	Start development mode
build.bat	Build project
start.bat	Start production server
start-tunnel.bat	Start with Cloudflare Tunnel
test.bat	Run all tests

macOS / Linux

npm run commands work identically on all platforms. Use the terminal instead of .bat scripts.

npm run dev        # Development mode
npm run build      # Build
npm run start      # Production server
npm test           # Run tests

Remote Access (Cloudflare Tunnel)

# Install cloudflared
winget install cloudflare.cloudflared    # Windows
brew install cloudflared                  # macOS

# Set TUNNEL_ENABLED=true in .env, then:
npm run start:tunnel
# → Outputs https://xxxx.trycloudflare.com in the console

Route a named tunnel through your own domain (optional)

To avoid the "dangerous site" browser warnings on *.trycloudflare.com / *.cfargotunnel.com, point a named tunnel at your own domain. Either use the sidebar ⚙ → Tunnel settings modal (Tunnel Name + Custom Hostname), or the CLI:

clitrigger config tunnel hostname app.your-domain.com
cloudflared tunnel route dns <tunnel-name> app.your-domain.com   # one-time

The displayed URL becomes https://app.your-domain.com and reputation tracks your domain.

---

Documentation

📖 The full manual lives in the Wiki — installation, every feature guide, and remote access.

Doc	Content
Wiki	Detailed feature guides and usage
SETUP.md	Detailed installation and usage guide (한국어)
changelog/	Version history (per-date entries by month)
CICD.md	GitHub Actions CI/CD setup
TESTING.md	Testing guide

---

Star & Join Us

If CLITrigger saves you time, please give us a star — it genuinely helps the project reach more developers.

Want to help shape what comes next? We're actively looking for contributors:

• File an issue — bug reports, feature requests, and rough ideas all welcome at Issues
• Open a PR — start with good first issue labels, or pick anything that itches you
• Share what you built — drop your worktree workflows, custom plugins, or productivity tips in Discussions

Every star, issue, and PR moves this faster. Thank you 🙏

---

Contributors

Thanks to everyone who has contributed to CLITrigger!

---

Star History

---

☕ Buy Me a Coffee

If CLITrigger saves you time, consider buying me a coffee!

---

License

MIT — Free to use, modify, and distribute.