Connecting to Contextium
Understand the two ways to connect external tools to your Contextium workspace — MCP for AI assistants and CLI for terminal workflows.
Contextium stores your team's knowledge — files, skills, agents, and workflows — in one place. Connecting your AI tools to Contextium means your assistant is grounded in that shared context every time it answers a question or performs a task. No more copy-pasting, no more each person feeding different context to their agent.
Beyond the web interface, Contextium offers two ways to connect: MCP and the CLI.
MCP server
MCP (Model Context Protocol) is an open standard that lets AI agents connect to external tools. When you add Contextium's MCP connection, your AI agents can browse your context libraries, search files, load skills and agents, run workflows, and manage tags — all from the chat interface.
This is the primary way to ground your AI assistant in your team's knowledge. Instead of manually providing context, the assistant pulls what it needs directly from your workspace.
Best for: Claude Desktop, VS Code Copilot, Cursor, or any tool that supports remote MCP. Or building custom agents in Microsoft Foundry or AWS Bedrock.
Setup: No installation required. You paste a short config snippet into your AI tool and authenticate in the browser.
CLI — for terminal workflows and agent pipelines
The Contextium CLI lets you browse, search, and manage your workspace from the terminal. It is also useful for feeding context into AI agents programmatically — for example, piping files and skills into a local assistive AI Agent, such as GitHub Coilot or Claude Code.
Best for: developers, CI/CD pipelines, bulk operations, scripting, and agent design workflows.
Requirements: Node.js 18 or later.