Autohand Code Integration
Connect Autohand Code’s built-in LLM Gateway provider with an API key, choose a model, and verify a coding task.
Autohand Code is a terminal coding agent with a built-in LLM Gateway provider. Use a pay-as-you-go or DevPass API key to connect it to your workspace.
Video walkthrough
Install
1pnpm add -g autohand-cli2autohand --version1pnpm add -g autohand-cli2autohand --versionSee the Autohand documentation for installation and onboarding.
Sign in to Autohand
The current CLI requires an Autohand account even when inference uses a provider API key. Run autohand login and complete the browser sign-in before starting a session. Your LLM Gateway key is a separate credential.
Connect LLM Gateway
Create a key in your dashboard, then export it before launching Autohand:
1export LLMGATEWAY_API_KEY="your_api_key"2autohand --provider llmgateway --model MODEL_ID1export LLMGATEWAY_API_KEY="your_api_key"2autohand --provider llmgateway --model MODEL_IDReplace MODEL_ID with a text model that supports tools from the live catalogue. With DevPass, choose a canonical model ID supported by your plan; provider-pinned routing is unavailable on coding plans.
Save the provider configuration
To keep LLM Gateway as your default, merge these settings into ~/.autohand/config.json:
1{2 "provider": "llmgateway",3 "llmgateway": {4 "apiKey": "${LLMGATEWAY_API_KEY}",5 "baseUrl": "https://api.llmgateway.io/v1",6 "model": "MODEL_ID"7 }8}1{2 "provider": "llmgateway",3 "llmgateway": {4 "apiKey": "${LLMGATEWAY_API_KEY}",5 "baseUrl": "https://api.llmgateway.io/v1",6 "model": "MODEL_ID"7 }8}The provider name is a string, with a separate llmgateway settings object. Keep the API key in your environment and replace the model placeholder before running the agent.
Verify a coding task
Start autohand from your project directory. Ask it to read a failing test, fix the implementation, and rerun the test without changing the test file. Review its file changes and command output, then check the request in the workspace that issued your key.
Troubleshooting
- Wrong provider: Pass
--provider llmgatewayexplicitly or check the top-levelprovidersetting. - Authentication failed: Export
LLMGATEWAY_API_KEYin the same shell and confirm the key is active. - Model unavailable: Choose a tool-capable model allowed by your workspace or DevPass plan.
See Autohand's LLM Gateway integration reference for additional options.