MCP Source Code

Unizo's Source Code MCP Server enables seamless access and management of source control systems through the Model Context Protocol (MCP). This extensible platform allows AI agents and LLM clients to interact with repositories, branches, pull requests, commits, and more across multiple source control systems.

Overview

Designed for AI-native developer workflows, this guide walks you through installation, toolset usage, and integration specifics for MCP-capable clients.

Key Benefits

• Unified SCM Interface: Interact with platforms like GitHub, GitLab, Bitbucket, and Azure DevOps through a single, consistent set of MCP tools
• AI Agent Integration: Compatible with Claude, GPT, and other LLMs supporting the MCP standard
• Extensive Coverage: Provides access to repositories, branches, commits, and pull requests with intelligent abstraction and normalization

Available Tools

Tool Reference

Tool Name	Description	Parameters
list_organizations	Retrieves organizations from a selected integration	Required: integration id/name Returns: Organization list with metadata
list_repositories	Lists repositories under a specific organization	Required: integration id/name, organization id/name Returns: Repository list with basic information
list_branches	Fetches branches within a given repository	Required: integration id/name, organization id/name, repository id/name Returns: Branch names and commit information
list_commits	Lists recent commits in a repository branch	Required: integration id/name, organization id/name, repository id/name, branch Returns: Commit history with messages and authors
list_pull_requests	Fetches pull requests from a repository	Required: integration id/name, organization id/name, repository id/name Returns: Pull request list with status and metadata
create_pull_request	Creates a new pull request	Required: integration id/name, organization id/name, repository id/name, title, sourceBranch, targetBranch Returns: Created pull request details and URL

Installation & Setup

Prerequisites

• Node.js or Python runtime (for the MCP proxy server)
• MCP-compatible agent (Claude, GPT, etc.)
• Unizo API credentials (see environment section)

Example MCP Server Setup (JSON)

{
  "mcpServers": {
    "unizo": {
      "command": "mcp-proxy",
      "args": ["http://172.184.130.51:8999/mcp"],
      "env": {
        "API_ACCESS_TOKEN": "your_account_token"
      }
    }
  }
}

tip

Use the same MCP orchestration layer as your other Unizo services to simplify deployment and logging.

Environment Variables

Variable	Description
API_ACCESS_TOKEN	Your Unizo Source Code MCP authentication token

Client Integration Flow

1. Start the MCP Server

   • Use your preferred orchestrator (e.g., Claude Desktop, GPT MCP Plugin)
   • Load the MCP JSON config with the unizo-scm target

2. Discover Tools

   • Use list_services, list_integrations, and list_organizations to explore available SCM integrations

3. Perform Repository Operations

   • Call list_repositories, list_branches, or list_pull_requests to explore content
   • Use get_file_content or create_pull_request to drive actions

4. Abstracted Control

   • The MCP server handles normalization across GitHub, GitLab, Bitbucket, and Azure DevOps—agents just pass in the structured context

Observability & Logging

All tool calls are logged by the Unizo Source Code MCP Server with agent attribution, request/response metadata, and correlation IDs for tracing and debugging.

Early Access Program

The Unizo Source Code MCP Server is currently available through our Early Access Program. All customers can gain priority access to these powerful identity management tools and receive dedicated implementation support.

Get Started Today

Reach out to our team to schedule a personalized demo and discuss integration options for your organization. Connect With Us