OpenCode Provider Bridge
Bring all your OpenCode-configured AI providers into VS Code's Chat model picker.
Any provider you've configured in OpenCode — Zen, Go, Anthropic, OpenAI, Google, NVIDIA, Vultr, and more — appears as a selectable model in the chat dropdown, ready to use alongside GitHub Copilot.
Quick Start
- Install the extension from the VS Code Marketplace
- Configure providers in OpenCode:
opencode /connect
- Open VS Code Chat → select a model from the dropdown
That's it. Providers are auto-discovered from a running opencode serve — the extension starts one automatically if needed.
Features
- All providers, one picker — Zen, Go, Anthropic, OpenAI, Google, and any other OpenCode provider
- Multi-SDK auto-routing — uses the correct AI SDK per model (
@ai-sdk/openai, @ai-sdk/openai-compatible, @ai-sdk/anthropic, @ai-sdk/google) based on opencode registry metadata
- Zero config — reads your existing OpenCode setup
- Secure keys — API keys stored in VS Code's encrypted SecretStorage
- Real-time reasoning — thinking content streams as it's generated
- Model configuration — per-model controls in the model picker: "Thinking Effort" (from opencode's per-model variants), "Context Size", "Temperature", and "Max Output Tokens"; selections are sent with each request
- Cost display — per-model cost shown in AI credits (converted from opencode's USD pricing)
- Server auto-start — starts
opencode serve headlessly when needed, with a retryable setup popup
- Token usage — status bar shows prompt/completion tokens per response
- Tool calling — full support for tools with schema simplification and cross-turn name resolution
- Debug logging — set
logLevel to debug to see raw SSE stream data, routing decisions, and request payloads
Requirements
- VS Code 1.120+
- OpenCode CLI installed and configured (
opencode /connect)
- A running opencode server is required — the extension auto-starts
opencode serve if none is running
Commands
| Command |
What it does |
OpenCode Bridge: Refresh Model List |
Re-discover providers and models |
OpenCode Bridge: Show Provider Status |
Show providers, key status, and model counts |
OpenCode Bridge: Set API Key for a Provider |
Set or change a provider's API key |
OpenCode Bridge: Remove Provider Key |
Remove a stored API key |
Extension Settings
| Setting |
Default |
Description |
opencode-provider-bridge.logLevel |
info |
Log verbosity: error, warn, info, or debug |
Source Files
| File |
Purpose |
extension.ts |
Extension entry point, BridgeProvider, provider discovery/caching |
serverManager.ts |
opencode server lifecycle (find/start, health check, retryable error popup) |
provider.ts |
OpencodeModelProvider — multi-SDK routing, message conversion, streaming, model config |
opencodeConfig.ts |
SDK-only provider/model discovery (mandatory opencode server) |
providerUtils.ts |
Schema simplification, tool result text extraction |
verboseFetch.ts |
SSE stream logging fetch wrapper (debug level) |
logger.ts |
Leveled logging with VS Code OutputChannel |
Changelog
See CHANGELOG.md for the full release history.
MIT
| |
