Use OpenCode with RunBridge AI

Connect OpenCode using its custom provider support and the OpenAI-compatible Chat Completions adapter.

Prerequisites

Install OpenCode from its official installation guide.
Create an active API key in the RunBridge AI dashboard.
Choose an exact model ID that supports Chat Completions, streaming, and tool calling. Replace your-model-id below.

The npm installation is:

npm install -g opencode-ai
opencode --version

Supply the API key

Bash / Zsh
PowerShell

printf 'RunBridge AI API key: '
read -rs RUNBRIDGE_API_KEY
printf '\n'
export RUNBRIDGE_API_KEY

$secureKey = Read-Host "RunBridge AI API key" -AsSecureString
$env:RUNBRIDGE_API_KEY = [System.Net.NetworkCredential]::new("", $secureKey).Password

Start OpenCode from this terminal so the process can read the variable.

Add a custom provider

Create opencode.json in your test project, or merge this into your existing configuration. For a configuration shared across your projects, use ~/.config/opencode/opencode.json.

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "runbridge": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "RunBridge AI",
      "options": {
        "baseURL": "https://api.runbridge.ai/v1",
        "apiKey": "{env:RUNBRIDGE_API_KEY}"
      },
      "models": {
        "your-model-id": {
          "name": "RunBridge AI coding model"
        }
      }
    }
  }
}

runbridge is a custom provider ID, not a built-in provider. The models key is the exact model ID sent to the API; name only controls its display label. Keep the base URL at /v1 because the adapter appends /chat/completions. OpenCode substitutes {env:RUNBRIDGE_API_KEY} at runtime. Keep that reference in the file instead of putting a secret into a project configuration. If you add context or output limits, use the chosen model’s documented values rather than copying another model’s limits.

This guide uses Chat Completions. For a Responses-only model, OpenCode documents the separate @ai-sdk/openai adapter. Select an adapter that matches the model’s API format before using it.

Verify the connection

Start OpenCode in the directory containing your configuration:

opencode

Run /models and select the RunBridge AI model. Ask: Reply with “RunBridge connected” without running commands or changing files. Check the request in the RunBridge AI usage records. Then verify tool calling in a disposable project before using the agent on real work.

Troubleshooting

Symptom	Check
Provider or model missing	Check JSON syntax, configuration location, and the custom provider/model names. Project settings can override global settings.
Missing key or 401	Export `RUNBRIDGE_API_KEY` in the terminal launching OpenCode and check the key in the dashboard.
404	Use `/v1` in `baseURL`, omit the operation path, and choose a Chat Completions-compatible model.
Unsupported tools or context-length error	Check the chosen model’s actual tool support and limits.

See the official custom provider and configuration documentation.

Get started

Coding agents

Agent platforms

Editor extensions

Use OpenCode with RunBridge AI

Prerequisites

Supply the API key

Add a custom provider

Verify the connection

Troubleshooting

​Prerequisites

​Supply the API key

​Add a custom provider

​Verify the connection

​Troubleshooting

Prerequisites

Supply the API key

Add a custom provider

Verify the connection

Troubleshooting