Skip to content

OpenCode Integration

OpenCode (and similar coding tools) can connect to AutoVio via the Model Context Protocol using the AutoVio MCP server.

Configure OpenCode to start the AutoVio MCP server as an external process, using the same entry point as other MCP clients:

Terminal window
node /path/to/AutoVio-MCP/dist/index.js \
--autovio-base-url http://localhost:3001 \
--autovio-api-token YOUR_API_TOKEN \
--vision-model gemini-2.0-flash-exp --vision-api-key YOUR_VISION_KEY \
--llm-model gemini-2.5-flash --llm-api-key YOUR_LLM_KEY \
--image-model gemini-2.5-flash-image --image-api-key YOUR_IMAGE_KEY \
--video-model veo-3.0-generate-001 --video-api-key YOUR_VIDEO_KEY

See MCP Setup for all supported flags and environment variables.

Alternatively, run the server with env variables:

Terminal window
AUTOVIO_BASE_URL=http://localhost:3001 \
AUTOVIO_API_TOKEN=your-token \
AUTOVIO_LLM_API_KEY=your-llm-key \
node /path/to/AutoVio-MCP/dist/index.js

You can add additional env variables for vision, image, and video models and keys as needed.

Once OpenCode connects to the MCP server, it can use the same tools documented in MCP Tools Reference, including:

  • autovio_projects_* — manage projects
  • autovio_works_* — manage works and run the pipeline
  • autovio_ai_* — direct AI operations
  • autovio_templates_* — manage templates
  • autovio_providers_list — discover providers and models

Refer to OpenCode’s MCP documentation for how to register an MCP server; the AutoVio side only requires the command and configuration described here.