LoopTroop Docs

Appearance

Getting Started ​

Welcome to LoopTroop! This guide will take you from zero to running your first AI-driven development cycle on your repository.

TIP

LoopTroop works best when it has a powerful "Council" of AI models to brainstorm ideas. You don't need a massive budget to get started—check out the Free AI Models section below!

1. Prerequisites ​

You need a few basic developer tools:

  • Node.js and npm
  • git
  • A local git repository that has an origin pointing to GitHub.
  • OpenCode installed locally — npm run dev will start and configure it automatically.

2. Setting Up Your AI Council (For Free!) ​

💸 Configuring Free AI Models ​

You no longer need to burn expensive tokens to run autonomous agents. LoopTroop supports all major API Gateways that provide state-of-the-art Mixture-of-Experts (MoE) models for free.

OpenRouter provides a unified API with a dynamic router that automatically selects available zero-cost models.

  1. Create a free account at https://openrouter.ai/.
  2. In your project's .env file set:
    bash
    PROVIDER=openrouter
OPENROUTER_API_KEY="your-api-key"
  3. Select the Model:
    • The Auto-Router: Set your model to openrouter/free. LoopTroop will automatically route tasks to available free models capable of tool-calling.
    • Specific High-Capacity Free Models:
      • nvidia/nemotron-3-super-120b-a12b:free (120B parameters, 1M context window)
      • qwen/qwen3-coder:free (480B parameters, best for heavy repository logic)
      • inclusionai/ling-2.6-flash:free (Extremely fast, highly token-efficient)
      • google/gemma-4-31b-it:free (Excellent for multimodal tasks)

Option B: NVIDIA NIM API Catalog ​

NVIDIA provides highly optimized, GPU-accelerated endpoints. Signing up gives you 1,000 base credits (up to 5,000 total trial credits).

  1. Go to build.nvidia.com and create a Developer account.
  2. Generate a personal key in the API Keys section.
  3. In your .env file:
    bash
    PROVIDER=nvidia-nim
NVIDIA_API_KEY="your-ngc-api-key"
  4. Recommended Free/Trial Models:
    • nemotron-3-super-120b-a12b
    • nemotron-3-nano-30b-a3b
    • mistral-small-4-119b-2603

Option C: OpenCode Free Network ​

OpenCode curates a validated list of models specifically benchmarked for agentic coding.

  1. Obtain your OpenCode API key from opencode.ai.

  2. In your .env file:

    bash
    PROVIDER=opencode
OPENCODE_API_KEY="your-opencode-key"

  3. Recommended Free Models:

    • big-pickle
    • nemotron-3-super-free
    • minimax-m2.5-free
    • mimo-v2-pro-free

    NOTE

    The OpenCode free-network model catalog evolves frequently. The names above were accurate at the time of writing but may change. Check the OpenCode model list or your ProfileSetup model picker (which loads the live catalog from your connected OpenCode server) for current names.

📊 Latency & Model Tracking Tools ​

Because free APIs can occasionally experience rate-limiting or latency spikes, the community maintains active trackers to help you route your agents efficiently:

  • free-ai-tools: The master directory of over 550 free APIs, IDEs, and local RAG stacks. Check this repository frequently for newly added free models and quota details.
  • ClawRouter: An open-source routing layer that tracks the real-time latency of top-tier free models and handles load balancing.
  • frouter: A fast CLI tool to ping free models and test Time To First Token (TTFT) before starting your loop.

Connecting OpenCode to OpenRouter ​

When running opencode serve, ensure your environment variables are set to use OpenRouter as your provider and supply your OpenRouter API key. See the OpenCode documentation for exact provider configuration.

3. Installation ​

Clone the LoopTroop repository and install the dependencies:

bash
git clone https://github.com/looptroop-ai/LoopTroop.git
cd LoopTroop
npm install

4. Starting the Application ​

The main development command starts the frontend, backend, docs, and the OpenCode watcher stack all at once. It also runs the normal startup maintenance checks for dependencies, npm audit fixes, and the local OpenCode CLI. npm run dev resolves the OpenCode server endpoint automatically — reusing a running local instance when available, or starting one on a free nearby port — and secures it with ephemeral basic auth credentials when a new local server is needed.

bash
npm run dev

For non-mutating startup, forced maintenance, verbose startup output, and manual maintenance commands, see Operations Guide.

By default, the services run on these ports:

ServiceAddress
Frontend (UI)http://localhost:5173
Backend (API)http://localhost:3000
Docshttp://localhost:5174
OpenCodehttp://127.0.0.1:4096

IMPORTANT

If OpenCode is running on a different port, you can tell LoopTroop where to find it using an environment variable: export LOOPTROOP_OPENCODE_BASE_URL=http://127.0.0.1:YOUR_PORT

5. Attaching Your First Project ​

Once the frontend is up:

  1. Open http://localhost:5173 in your browser.
  2. Click Add Project and provide the absolute path to your local git repository.
  3. LoopTroop will verify that it is a valid git repository with a GitHub origin.
  4. You're ready to create your first Ticket!

Operations and Troubleshooting ​

For runtime storage, environment variables, startup maintenance, disk cleanup, diagnostics, and OpenCode troubleshooting, see Operations Guide.

Next Steps ​

To dive deeper into how LoopTroop actually plans and executes your code, check out:

Morty Proxy This is a proxified and sanitized view of the page, visit original site.