Valossa Assistant MCP Server Setup

The Valossa Video AI Assistant (Valossa Assistant) can be integrated into agentic ecosystems via the Model Context Protocol (MCP). This allows LLM orchestrators, developer coding environments (such as Cursor or VS Code), and native desktop or CLI AI clients (such as Claude Code and Claude Desktop) to parse, query, and interpret your video libraries directly. In short, all this is the answer to the question: "How to use AI agents to analyze video?"

The Valossa Assistant MCP server implements the production-grade streamable-http transport protocol and native OAuth 2.0 user authentication. Instead of handling raw audio-visual streams inside the LLM context window, your AI agents safely interface with Valossa's 7th-generation multimodal AI infrastructure using explicit permission scopes.

Detailed documentation of the available Valossa Assistant MCP tools is provided in the Valossa Assistant Canonical MCP Tool Documentation.

To use the MCP server, you first need to create a Valossa Assistant user account. Usage is tracked against your account quota regardless of the interaction channel: Valossa Assistant can be used as an online app via a web browser or as an agentic toolbox via the MCP connection.

Connection Specifications

To configure your client application, connect using the following production endpoint and protocol variables:

• Protocol Transport: streamable-http / JSON-RPC 2.0
• Endpoint Base URL: https://mcp-assistant.valossa.com/mcp
• Authentication Method: OAuth 2.0 (Launches browser authentication in the Valossa Assistant online app automatically on first connection)
• Discovery Metadata: Available at https://assistant.valossa.com/.well-known/mcp.json and https://mcp-assistant.valossa.com/.well-known/oauth-protected-resource

The items above are a simplification. For more technical details, please refer to the Valossa Assistant Canonical MCP Documentation page.

Client Configuration

For your convenience, we repeat below the basic setup instructions, which can also be found on the canonical documentation page of the Valossa Assistant MCP server. Please follow the instructions applicable to your specific execution environment to integrate the MCP server correctly.

1. Claude Code (CLI)

To add the Valossa Assistant MCP server natively to your Claude CLI agent workspace, run the following command directly in your terminal:

claude mcp add --transport http valossa https://mcp-assistant.valossa.com/mcp

2. Claude Desktop, Cursor, and JSON-Config Clients

Add an entry for the Valossa Assistant MCP server inside your local application configuration profile (e.g., claude_desktop_config.json):

{
  "mcpServers": {
    "valossa": {
      "type": "http",
      "url": "https://mcp-assistant.valossa.com/mcp"
    }
  }
}

(Note: Some JSON clients call the key servers instead of mcpServers, or the field transport instead of type. Check your specific client's documentation. The URL and transport configurations remain identical everywhere).

Tool Specifications and Reference

The server splits capabilities into distinct operational categories. Read-Only tools execute instantly without explicit user confirmation prompts, while Writes Data tools (such as file uploads or clip exports) consume your account quota credits and prompt for explicit execution permission before running.

📖 View Live Tool Documentation & Interactive Examples

For the absolute latest human-readable breakdown of our exact parameters, schema constraints, and example JSON payloads, visit our auto-rendered production directory: 👉 Valossa Assistant Canonical MCP Tool Documentation

Integration Constraints & Guardrails

⚠️ Asynchronous Multi-Tier Latency

Processing operations divide tracking components dynamically onto parallel operational pipelines:

1. Speech Logs: Typically evaluate faster than the video playback runtime. Transcripts and dialogue arrays become active and queryable within moments.
2. Computer Vision & Themes: In-depth image parsing metrics (objects, faces, human expression patterns, custom branding checks) typically will compile completely slightly after the aggregate file play timeline length.

🔒 Enterprise Sovereignty & Data Compliance

All streaming payload transfers conform fully to GDPR and the EU AI Act:

• Backend nodes remain entirely within secure European cloud computing zones.
• Video chunks, processing logs, and interaction transcripts are explicitly never introduced to train foundational machine learning systems or public baseline sets.

Frequently Asked Questions (FAQ)

What is the Valossa Model Context Protocol (MCP) server?

The Valossa MCP server is an enterprise remote gateway that allows Large Language Models (LLMs) and AI agents to seamlessly watch and query video libraries. By utilizing the official streamable-http transport standard and OAuth 2.0 client authentication, it allows agents to perform deep semantic search, retrieve accurate frame logs, and generate clips safely.

How does the server handle user security and authentication?

Valossa utilizes industry-standard OAuth 2.0 protocols mapped against your native Valossa Assistant user account. Your password remains hidden from connected client software frameworks. Upon launching connection configurations for the first time, your application environment safely spawns a secure browser-based login prompt page to generate local short-lived authorization tokens automatically.

Can an AI agent directly splice or clip video assets?

Yes. By utilizing the tools made available in the Valossa Assistant MCP specification, an LLM agent can parse a timeline via simple conversational queries, request targeted highlights, and command the compiler service to deliver (for example) downloadable .mp4 video files or accurate .srt caption files.

Where can I check live updates to input schemas or argument structures?

While this guide provides general installation configurations and client environment onboarding examples, the live, programmatic description database tracking every single operational rule can be verified on our active production endpoint directory at mcp-assistant.valossa.com/documentation.