Uniquity Mcp Server

@KunihiroS

About Uniquity Mcp Server

Uniquity MCP Server is a server that makes the functions of UniquityReporter available to external tools and AI agents via MCP (Model Context Protocol).

#uniquity #mcp #server

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "uniquity-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "uniquity-mcp@latest"
      ],
      "env": {
        "GITHUB_TOKEN": "{apikey}",
        "OPENAI_API_KEY": "{apikey}",
        "TAVILY_API_KEY": "{apikey}"
      }
    }
  }
}

Tools

Analyzes a Git repository and generates a report using Uniquity Reporter. The analysis is performed with repo=off mode, meaning no local repository copy is created or persisted.

Overview

What is Uniquity MCP Server?

Uniquity MCP Server makes the analysis functions of UniquityReporter available to external tools and AI agents via the Model Context Protocol (MCP). It separates core logic from interface, enabling standardized access from MCP hosts and clients.

How to use Uniquity MCP Server?

Configure the MCP host with the required environment variables (GITHUB_TOKEN, OPENAI_API_KEY, TAVILY_API_KEY) and run via npx -y uniquity-mcp@latest. Use the provided tools (analyze_repository, list_tools) to generate similarity analysis reports for GitHub repositories in Markdown format.

Key features of Uniquity MCP Server

Generates similarity analysis reports for GitHub repositories.
Returns reports in Markdown format via standard output.
Supports dynamic selection of OpenAI models and log output.
Provides a list_tools endpoint for discovering available tools.
Relies on UniquityReporter as its core analysis engine.

Use cases of Uniquity MCP Server

AI agents analyzing GitHub repository similarity on demand.
MCP host applications integrating repository analysis into workflows.
Automation of code similarity checks across multiple repositories.
Dynamic report generation with configurable OpenAI models.

FAQ from Uniquity MCP Server

What is UniquityReporter?

UniquityReporter is the core analysis engine that Uniquity MCP Server depends on to perform repository analysis, code analysis, and web search through external services (GitHub API, OpenAI API, Tavily API).

What environment variables are required?

GITHUB_TOKEN, OPENAI_API_KEY, and TAVILY_API_KEY must be set. Optional variables include openaiModel and logEnabled (defaults from config/config.js apply if omitted).

Does Uniquity MCP Server save reports to files?

No, reports and logs are always output to standard output and are not saved to files. If file saving is needed, the host should capture the standard output.

What tools does Uniquity MCP Server provide?

analyze_repository – generates a similarity analysis report for a specified GitHub repository. list_tools – returns a list of all available tools with their specifications.

Can I specify the OpenAI model dynamically?

Yes, the analyze_repository tool accepts an optional openaiModel argument to override the default model (e.g., o3-mini, gpt-4.1-nano). The logEnabled argument can also be set per call.

Frequently asked questions

What is UniquityReporter?

What environment variables are required?

`GITHUB_TOKEN`, `OPENAI_API_KEY`, and `TAVILY_API_KEY` must be set. Optional variables include `openaiModel` and `logEnabled` (defaults from `config/config.js` apply if omitted).

Does Uniquity MCP Server save reports to files?

No, reports and logs are always output to standard output and are not saved to files. If file saving is needed, the host should capture the standard output.

What tools does Uniquity MCP Server provide?

`analyze_repository` – generates a similarity analysis report for a specified GitHub repository. `list_tools` – returns a list of all available tools with their specifications.

Can I specify the OpenAI model dynamically?

Yes, the `analyze_repository` tool accepts an optional `openaiModel` argument to override the default model (e.g., o3-mini, gpt-4.1-nano). The `logEnabled` argument can also be set per call.