Skip to content

MCP Overview

The AutoVio MCP server lets you call the AutoVio API directly from MCP-compatible clients such as Claude Desktop. It exposes tools for:

  • Health checks
  • Authentication (login/register/me)
  • Project and work management
  • AI analysis, scenario, image and video generation
  • Provider and template management

Instead of writing HTTP requests by hand, you call MCP tools like autovio_projects_create or autovio_ai_generate_image from your AI assistant.

  • The MCP server is a Node.js process that:
    • Loads configuration (AutoVio base URL, API token, AI models and keys).
    • Creates an AutoVioClient to talk to the AutoVio REST API.
    • Registers tools with the MCP SDK.
    • Communicates with the MCP client over stdio.
  • Clients (e.g. Claude Desktop) start the server with "command": "node", "args": [".../dist/index.js", ...] in their MCP config.

When you call a tool in the MCP client, the server:

  1. Validates the tool input with Zod schemas.
  2. Calls the corresponding AutoVio API endpoint(s).
  3. Returns formatted JSON text back to the client.

From the MCP client’s perspective, tools are grouped roughly as:

GroupExamplesDescription
Healthautovio_healthCheck API and MCP connectivity.
Authautovio_auth_login, autovio_auth_register, autovio_auth_meAuthenticate and fetch current user info.
Projectsautovio_projects_list, autovio_projects_create, autovio_projects_get, autovio_projects_update, autovio_projects_deleteManage projects.
Worksautovio_works_list, autovio_works_create, autovio_works_get, autovio_works_update, autovio_works_delete, autovio_works_apply_template, autovio_works_generate_scenario, autovio_works_generate_sceneManage works and run the video pipeline.
AIautovio_ai_analyze_video, autovio_ai_generate_scenario, autovio_ai_generate_image, autovio_ai_generate_videoDirect AI operations not tied to a specific work.
Providersautovio_providers_listList AI providers and models.
Templatesautovio_templates_list, autovio_templates_get, autovio_templates_create, autovio_templates_update, autovio_templates_deleteManage editor templates.

See MCP Tools Reference for full input schemas and usage examples.

Use MCP when:

  • You are inside an AI IDE or assistant (Claude Desktop, Cursor, etc.).
  • You want the assistant to call tools directly without managing HTTP and auth.

Use the REST API when:

  • You integrate AutoVio into your own backend services or automation workflows.
  • You want full control over HTTP requests and responses.
  • Node.js >= 18
  • Access to an AutoVio API instance (base URL + API token or credentials)