Quick start
You need Python 3.13–3.15 and an OpenAI-compatible endpoint, API key, and available model name.
1. Install
For Linux/macOS:
git clone https://github.com/nl8590687/lemonclaw.git
cd lemonclaw
python -m venv .venv
source .venv/bin/activate
python -m pip install -e . -c constraints.txt
cp .lemonclaw/.env.example .lemonclaw/.env
In Windows PowerShell, activate with .venv\Scripts\Activate.ps1 and copy the config with Copy-Item .lemonclaw/.env.example .lemonclaw/.env. Other installation steps are the same.
2. Configure your model
Edit .lemonclaw/.env using your provider's values. These are placeholders, not a working endpoint:
OPENAI_BASE_URL=https://your-provider.example/v1
OPENAI_API_KEY=your-api-key
MODEL_NAME=your-model-name
ENABLE_WEBHOOK=false
ENABLE_FEISHU=false
Verify the terminal first, then enable additional channels as needed. The shell tool is off by default.
3. Start and verify
lemonclaw
Keep the project root as your working directory. At the terminal prompt, ask “Hello, what can you do?” A reply verifies the model connection and basic conversation. Use /help for commands and /exit to quit.
If the first request fails, check the endpoint, model name, API key, and network; see troubleshooting.