Connect your client

Any client that speaks OpenAI-compatible Chat Completions works with Boardwalk: coding agents, gateways, and chat interfaces. Three settings connect all of them — a base URL, a key, and a model ID — and the sections below give the exact configuration for the clients people ask about most. Use a model suited to your workload, with tool calling enabled when your agent needs it.

Connection settings

Custom providers do not always discover model limits. Set them explicitly when your client requires it, and reserve context for the answer.

OpenCode

Add a custom provider to opencode.json. Replace support-agent with your model ID.

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "boardwalk": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Boardwalk",
      "options": {
        "baseURL": "https://api.boardwalk.cloud/v1",
        "apiKey": "{env:BOARDWALK_API_KEY}"
      },
      "models": {
        "support-agent": { "name": "Support agent" }
      }
    }
  },
  "model": "boardwalk/support-agent"
}

Cline, Roo Code and Kilo Code

All three are the same VS Code lineage and connect the same way. In the extension's settings, choose the OpenAI Compatible provider and fill in:

Base URL:  https://api.boardwalk.cloud/v1
API Key:   $BOARDWALK_API_KEY
Model ID:  support-agent

These clients do not read a model's limits from the provider, so set the context window and max output yourself from the model's published configuration, and enable the tool-calling toggle if the extension offers one.

Continue

Continue reaches any OpenAI-compatible endpoint through its built-in openai provider, so no plugin is needed — only an apiBase.

models:
  - name: Boardwalk
    provider: openai
    model: support-agent
    apiKey: ${{ secrets.BOARDWALK_API_KEY }}
    apiBase: https://api.boardwalk.cloud/v1
    roles: [chat, edit, apply]
    capabilities: [tool_use]

Crush

Add Boardwalk as a provider in crush.json. Declare each model's limits: Crush uses them to decide how much context it can spend.

{
  "$schema": "https://charm.land/crush.json",
  "providers": {
    "boardwalk": {
      "type": "openai",
      "base_url": "https://api.boardwalk.cloud/v1",
      "api_key": "$BOARDWALK_API_KEY",
      "models": [
        {
          "id": "support-agent",
          "name": "Support agent",
          "context_window": 32768,
          "default_max_tokens": 8192
        }
      ]
    }
  }
}

Aider

Set the OpenAI-compatible connection and prefix the model ID with openai/.

export OPENAI_API_BASE="https://api.boardwalk.cloud/v1"
export OPENAI_API_KEY="$BOARDWALK_API_KEY"
aider --model openai/support-agent

LiteLLM

Route through a LiteLLM proxy by declaring Boardwalk as an OpenAI-compatible upstream. Any tool that speaks to LiteLLM then reaches Boardwalk.

model_list:
  - model_name: support-agent
    litellm_params:
      model: openai/support-agent
      api_base: https://api.boardwalk.cloud/v1
      api_key: os.environ/BOARDWALK_API_KEY

Other clients

Open WebUI, LibreChat, Jan, Zed and anything else with an OpenAI-compatible provider take the same three settings. Check which API the client sends requests to: a configurable base URL alone does not establish compatibility with /v1/chat/completions.

Before a long run

• Confirm the model's verified capabilitiesmatch your client, then send a short test request.
• Set a suitable timeout for model startup and long responses.
• Check your credit balance or enable auto top-up.
• Use receipts as the authoritative cost record; client estimates can differ.