proq
agent
Fail
Health Warn
- No license — Repository has no license file
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 15 GitHub stars
Code Fail
- os.homedir — User home directory access in desktop/src/main/config.ts
- process.env — Environment variable access in desktop/src/main/config.ts
- process.env — Environment variable access in desktop/src/main/index.ts
- execSync — Synchronous shell command execution in desktop/src/main/server.ts
- process.env — Environment variable access in desktop/src/main/server.ts
Permissions Pass
- Permissions — No dangerous permissions requested
Purpose
This is a local, kanban-style task management agent for Claude Code instances. It allows developers to spawn parallel AI agents, giving each task its own isolated git worktree to work on the same codebase simultaneously without conflicts.
Security Assessment
The overall risk is Medium. The tool actively accesses the user's home directory and reads environment variables to configure its local server. More critically, it uses `execSync` (synchronous shell command execution) in its server file. While this is expected given its core function of automating local git worktrees and executing agent commands, it fundamentally introduces remote/code-driven command execution on your machine. There are no hardcoded secrets detected, and the application operates entirely locally without requesting dangerous system permissions.
Quality Assessment
The project appears to be actively maintained, with repository pushes happening as recently as today. It has a modest but present community backing with 15 GitHub stars. The repository includes extensive documentation covering architecture, worktrees, and setup. It properly lists an MIT license in the README, though the automated scanner failed to find a dedicated license file in the repository root.
Verdict
Use with caution — while it is an actively maintained local tool, its requirement to execute shell commands and access system paths means you should review the exact code being run before allowing it to manage your local development environment.
This is a local, kanban-style task management agent for Claude Code instances. It allows developers to spawn parallel AI agents, giving each task its own isolated git worktree to work on the same codebase simultaneously without conflicts.
Security Assessment
The overall risk is Medium. The tool actively accesses the user's home directory and reads environment variables to configure its local server. More critically, it uses `execSync` (synchronous shell command execution) in its server file. While this is expected given its core function of automating local git worktrees and executing agent commands, it fundamentally introduces remote/code-driven command execution on your machine. There are no hardcoded secrets detected, and the application operates entirely locally without requesting dangerous system permissions.
Quality Assessment
The project appears to be actively maintained, with repository pushes happening as recently as today. It has a modest but present community backing with 15 GitHub stars. The repository includes extensive documentation covering architecture, worktrees, and setup. It properly lists an MIT license in the README, though the automated scanner failed to find a dedicated license file in the repository root.
Verdict
Use with caution — while it is an actively maintained local tool, its requirement to execute shell commands and access system paths means you should review the exact code being run before allowing it to manage your local development environment.
A task board for Claude. Precision vibe coding tool.
README.md
Serious vibe coding
Download · Run Locally · Docs
Proq is an agentic development environment for serious vibe coding.
A kanban-style task manager for local Claude Code instances.
Create a task → proq spins up an agent, gives it an isolated worktree, watch the agent work (or not), and preview changes.
The point is maintaining quality and clarity when vibe coding. Automatic history, fresh focused contexts, and parallel ops. It's twice as effective as a screen full of terminals.
Free, no signup, local-only, and works with whatever agent config, MCPs, and tools you already use with Claude.
Download
macOS app — get moving in 1 minute
| Platform | Link |
|---|---|
| macOS (Apple Silicon) | Download .dmg |
Just open it and start cooking.
Run Locally
Requires Node.js 18+ and the Claude Code CLI on your PATH.
git clone https://github.com/0xc00010ff/proq.git
cd proq
npm run setup
npm run dev
Open localhost:1337. Add proq as a project. Create a task and watch the app update itself.
Explore
- Parallel agents — each task gets its own git worktree and branch, multiple agents work the same codebase without conflicts
- Live preview — start and view your project live, let agents view/use your app
- Project workbench — freeform agent, terminal, and code editor for quick little edits
- HTTP API — every board action is a REST endpoint, anything that can make HTTP requests can manage tasks
- MCP server — manage projects and tasks from any MCP-compatible agent or tool
- Supervisor — an agent that lives above all your projects, can be hooked up to OpenClaw / external agents
- Customization - toggle kanban vs list, pretty chat vs raw CLI, light/dark mode
- Self-editing — add proq to its own project list, add whatever features you want
Docs
| Doc | What it covers |
|---|---|
| Getting Started | Install, create tasks, watch agents, review |
| Architecture | Data layer, dispatch engine, API routes, settings |
| Parallel Worktrees | Worktree lifecycle, preview branches, merge conflicts |
| Self-Editing | Developing proq with proq |
| Desktop App | Electron app setup and packaging |
License
MIT
proq was built using proq
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found