setting-up-oh-my-opencode

Install oh-my-opencode using npx with platform-specific flags: on macOS, run `npx oh-my-opencode install --no-tui --claude=yes --openai=yes --gemini=yes`; on devbox, use `--claude=max20` instead. The installer configures model routing and updates ~/.config/opencode/oh-my-opencode.json with your credentials and model mappings.

Yes, oh-my-opencode routes Claude, Gemini, OpenAI, and Vertex AI simultaneously. The plugin maps each model in ~/.config/opencode/oh-my-opencode.json, enabling you to switch between providers and access them via the opencode CLI without reconfiguration.

Authentication setup is included in the oh-my-opencode installer; it prompts for API credentials during installation. After setup, verify access by running `opencode models | grep google-vertex` or test a model call with `opencode run --model google-vertex-anthropic/claude-opus-4-5@20251101`.

Yes, oh-my-opencode supports both macOS and devbox platforms. Installation flags differ slightly per platform—macOS uses `--claude=yes` while devbox accepts `--claude=max20`—but both complete the same configuration and credential mapping process.

After installing oh-my-opencode, verify model access by running `opencode models | grep google-vertex` to list available Vertex AI models, or execute a test command like `opencode run --model google-vertex-anthropic/claude-opus-4-5@20251101 "Say 'Vertex AI working'"` to confirm routing is active.

If routing fails, check that ~/.config/opencode/oh-my-opencode.json contains valid API credentials and correct model mappings for each provider. Re-run the installer or manually update the configuration file, then verify with `opencode models` to confirm all providers are registered.