Experimental · manual fallback

OpenCode + Gruvo

OpenCode 1.x provider using its documented OpenAI-compatible config and native /connect credential flow.

>=1.0.0 <2.0.0native promptchat completionsstreamingtools

Experimental: A real OpenCode streaming tool-driven repository-edit conformance run has not yet been recorded.

1
Check your version
Run opencode debug paths config and confirm it satisfies >=1.0.0 <2.0.0.
2
Add the Gruvo provider
Update only the declared targets below, pointing provider traffic at https://inference-api.gruvo.xyz/v1. Preserve all unrelated settings.
3
Enter your key locally
Use native prompt so the key stays in the agent's native store or environment — never this website.
4
Verify
Run opencode debug paths config, opencode debug config, opencode auth list, opencode models gruvo and confirm the Gruvo models are visible. Your existing default model is preserved.

What gets written

~/.config/opencode/opencode.json

jsonc

{
  "provider": {
    "gruvo": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Gruvo",
      "options": {
        "baseURL": "https://inference-api.gruvo.xyz/v1"
      },
      "models": {
        "<available-model-id>": {
          "name": "<display-name>"
        }
      }
    }
  }
}

Values in <angle-brackets> come from your Gruvo project at setup time. Secret values are always redacted.

Reload OpenCode after approval if the provider list is cached.

Official OpenCode documentation