Cline

Set up Cline’s OpenAI-compatible provider, model capabilities, and separate Plan/Act configuration.

Install, configure, and test

1. In VS Code Extensions, search for Cline and install it; the official installation instructions are at https://docs.cline.bot/getting-started/installing-cline. Open the Cline sidebar and its gear icon. In the routerra.ai console (https://routerra.ai/keys), create a gateway API key and copy a model ID accessible to that key.

2. Set API Provider to OpenAI Compatible, Base URL to https://routerra.ai/v1, API Key to YOUR_API_KEY, and Model ID to YOUR_MODEL_ID (some versions label this Model). The base excludes /chat/completions: Cline builds the operation path. Leave Use Azure Identity Authentication off; this setup authenticates with the gateway key, not Azure or a Cline subscription.

3. Expand Model Configuration. Set Context Window size and Max Output Tokens to the selected model's actual limits. Enable Image Support only for a vision model and Computer Use only when that model and route support it. Local input/output prices describe Cline's estimate; they do not override gateway billing. If using different Plan and Act models, configure both modes separately.

4. Save the settings, reload the VS Code window if the old provider persists, and start a fresh task asking Cline to read a small workspace file. Check streamed output and the tool result, then inspect the gateway log.

For model-not-found errors, compare Model ID with the console. A 404 can mean a full operation URL was pasted into Base URL. If images or long tasks fail, correct the capability flags and context/output limits rather than merely increasing every value.