productivitydeveloper-tools

Probe Documentation Search

by probelabs

Search any codebase or documentation, including Git Hub repositories, with Probe's optimized, auto-updating search engin

Makes any codebase or documentation searchable through the Probe search engine with support for both local directories and Git repositories, featuring automatic content updates and build-time optimization that removes binary files to focus on searchable text content.

github stars

85

Works with any Git repositoryPowered by Probe search engineZero setup with npx command

best for

  • / Developers exploring unfamiliar codebases
  • / Teams wanting to chat with their documentation
  • / Building custom MCP servers for specific projects

capabilities

  • / Search any GitHub repository or codebase
  • / Query documentation from local directories
  • / Auto-update content from Git repositories
  • / Filter out binary files to focus on searchable text
  • / Pre-build documentation into packages
  • / Configure search tools dynamically at runtime

what it does

Makes any documentation or codebase searchable by AI assistants using the Probe search engine. Point it to a Git repository or local folder to enable natural language queries about code and docs.

about

Probe Documentation Search is a community-built MCP server published by probelabs that provides AI assistants with tools and capabilities via the Model Context Protocol. Search any codebase or documentation, including Git Hub repositories, with Probe's optimized, auto-updating search engin It is categorized under productivity, developer tools. This server exposes 1 tool that AI clients can invoke during conversations and coding sessions.

how to install

You can install Probe Documentation Search in your AI client of choice. Use the install panel on this page to get one-click setup for Cursor, Claude Desktop, VS Code, and other MCP-compatible clients. This server runs locally on your machine via the stdio transport.

license

MIT

Probe Documentation Search is released under the MIT license. This is a permissive open-source license, meaning you can freely use, modify, and distribute the software.

readme

Docs MCP Server

A flexible Model Context Protocol (MCP) server powered by Probe that makes any documentation or codebase searchable by AI assistants.

Chat with code or your docs by simply pointing to a git repo or folder:

npx -y @probelabs/docs-mcp@latest --gitUrl https://github.com/probelabs/probe

Use Cases:

  • Chat with any GitHub Repository: Point the server to a public or private Git repository to enable natural language queries about its contents.
  • Search Your Documentation: Integrate your project's documentation (from a local directory or Git) for easy searching.
  • Build Custom MCP Servers: Use this project as a template to create your own official MCP servers tailored to specific documentation sets or even codebases.

The content source (documentation or code) can be pre-built into the package during the npm run build step, or configured dynamically at runtime using local directories or Git repositories. By default, when using a gitUrl without enabling auto-updates, the server downloads a .tar.gz archive for faster startup. Full Git cloning is used only when autoUpdateInterval is greater than 0.

Features

  • Powered by Probe: Leverages the Probe search engine for efficient and relevant results.
  • Flexible Content Sources: Include a specific local directory or clone a Git repository.
  • Pre-build Content: Optionally bundle documentation/code content directly into the package.
  • Dynamic Configuration: Configure content sources, Git settings, and MCP tool details via config file, CLI arguments, or environment variables.
  • Automatic Git Updates: Keep content fresh by automatically pulling changes from a Git repository at a configurable interval.
  • Customizable MCP Tool: Define the name and description of the search tool exposed to AI assistants.
  • AI Integration: Seamlessly integrates with AI assistants supporting the Model Context Protocol (MCP).

Installation

Quick Start with Claude Desktop

Add to your Claude Desktop configuration file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "docs-search": {
      "command": "npx",
      "args": [
        "-y",
        "@probelabs/docs-mcp@latest",
        "--gitUrl",
        "https://github.com/your-org/your-repo",
        "--toolName",
        "search_docs",
        "--toolDescription",
        "Search documentation"
      ]
    }
  }
}

MCP Client Integration

You can configure your MCP client to launch this server using npx. Here are examples of how you might configure a client (syntax may vary based on the specific client):

Example 1: Dynamically Searching a Git Repository (Tyk Docs)

This configuration tells the client to run the latest @probelabs/docs-mcp package using npx, pointing it dynamically to the Tyk documentation repository. The -y argument automatically confirms the npx installation prompt. The --toolName and --toolDescription arguments customize how the search tool appears to the AI assistant.

{
  "mcpServers": {
    "tyk-docs-search": {
      "command": "npx",
      "args": [
        "-y",
        "@probelabs/docs-mcp@latest",
        "--gitUrl",
        "https://github.com/TykTechnologies/tyk-docs",
        "--toolName",
        "search_tyk_docs",
        "--toolDescription",
        "Search Tyk API Management Documentation"
      ],
      "enabled": true
    }
  }
}

Alternatively, some clients might allow specifying the full command directly. You could achieve the same as Example 1 using:

npx -y @probelabs/docs-mcp@latest --gitUrl https://github.com/TykTechnologies/tyk-docs --toolName search_tyk_docs --toolDescription "Search Tyk API Management Documentation"

Example 2: Using a Pre-built, Branded MCP Server (e.g., Tyk Package)

If a team publishes a pre-built package containing specific documentation (like @tyk-technologies/docs-mcp), the configuration becomes simpler as the content source and tool details are baked into that package. The -y argument is still recommended for npx.

{
  "mcpServers": {
    "tyk-official-docs": {
      "command": "npx",
      "args": [
        "-y",
        "@tyk-technologies/docs-mcp@latest"
      ],
      "enabled": true
    }
  }
}

This approach is ideal for distributing standardized search experiences for official documentation or codebases. See the "Creating Your Own Pre-built MCP Server" section below.

Here is example on how Tyk team have build own documentation MCP server https://github.com/TykTechnologies/docs-mcp.

Configuration

Create a docs-mcp.config.json file in the root directory to define the default content source and MCP tool details used during the build and at runtime (unless overridden by CLI arguments or environment variables).

Example 1: Using a Local Directory

{
  "includeDir": "/Users/username/projects/my-project/docs",
  "toolName": "search_my_project_docs",
  "toolDescription": "Search the documentation for My Project.",
  "ignorePatterns": [
    "node_modules",
    ".git",
    "build",
    "*.log"
  ]
}

Example 2: Using a Git Repository

{
  "gitUrl": "https://github.com/your-org/your-codebase.git",
  "gitRef": "develop",
  "autoUpdateInterval": 15,
  "toolName": "search_codebase",
  "toolDescription": "Search the main company codebase.",
  "ignorePatterns": [
    "*.test.js",
    "dist/",
    "__snapshots__"
  ]
}

Configuration Options

  • includeDir: (Build/Runtime) Absolute path to a local directory whose contents will be copied to the data directory during build, or used directly at runtime if dataDir is not specified. Use this OR gitUrl.
  • gitUrl: (Build/Runtime) URL of the Git repository. Use this OR includeDir.
    • If autoUpdateInterval is 0 (default), the server attempts to download a .tar.gz archive directly (currently assumes GitHub URL structure: https://github.com/{owner}/{repo}/archive/{ref}.tar.gz). This is faster but doesn't support updates.
    • If autoUpdateInterval > 0, the server performs a git clone and enables periodic updates.
  • gitRef: (Build/Runtime) The branch, tag, or commit hash to use from the gitUrl (default: main). Used for both tarball download and Git clone/pull.
  • autoUpdateInterval: (Runtime) Interval in minutes to automatically check for Git updates (default: 0, meaning disabled). Setting this to a value > 0 enables Git cloning and periodic git pull operations. Requires the git command to be available in the system path.
  • dataDir: (Runtime) Path to the directory containing the content to be searched at runtime. Overrides content sourced from includeDir or gitUrl defined in the config file or built into the package. Useful for pointing the server to live data without rebuilding.
  • toolName: (Build/Runtime) The name of the MCP tool exposed by the server (default: search_docs). Choose a descriptive name relevant to the content.
  • toolDescription: (Build/Runtime) The description of the MCP tool shown to AI assistants (default: "Search documentation using the probe search engine.").
  • ignorePatterns: (Build/Runtime) An array of glob patterns.
  • enableBuildCleanup: (Build) If true (default), removes common binary/media files (images, videos, archives, etc.) and files larger than 100KB from the data directory after the build step. Set to false to disable this cleanup.
    • If using includeDir during build: Files matching these patterns are excluded when copying to data. .gitignore rules are also respected.
    • If using gitUrl or dataDir at runtime: Files matching these patterns within the data directory are ignored by the search indexer.

Precedence:

  1. Runtime Configuration (Highest): CLI arguments (--dataDir, --gitUrl, etc.) and Environment Variables (DATA_DIR, GIT_URL, etc.) override all other settings. CLI arguments take precedence over Environment Variables.
  2. Build-time Configuration: Settings in docs-mcp.config.json (includeDir, gitUrl, toolName, etc.) define defaults used during npm run build and also serve as runtime defaults if not overridden.
  3. Default Values (Lowest): Internal defaults are used if no configuration is provided (e.g., toolName: 'search_docs', autoUpdateInterval: 5).

Note: If both includeDir and gitUrl are provided in the same configuration source (e.g., both in the config file, or both as CLI args), gitUrl takes precedence.

Creating Your Own Pre-built MCP Server

You can use this project as a template to create and publish your own npm package with documentation or code pre-built into it. This provides a zero-configuration experience for users (like Example 2 above).

  1. Fork/Clone this Repository: Start with this project's code.
  2. Configure docs-mcp.config.json: Define the includeDir or gitUrl pointing to your content source. Set the default toolName and toolDescription.
  3. Update package.json: Change the name (e.g., @my-org/my-docs-mcp), version, description, etc.
  4. Build: Run npm run build. This clones/copies your content into the data directory and makes the package ready.
  5. Publish: Run npm publish (you'll need npm authentication configured).

Now, users can run your specific documentation server easily: npx @my-org/my-docs-mcp@latest.

(The previous "Running", "Dynamic Configuration at Runtime", and "Environment Variables" sections have been removed as npx usage with arguments within client configurations is now the primary documented method.)

Using with AI Assistants

This MCP server exposes a search tool to connected AI assistants via the Model Context Protocol. The tool's name and description are configurable (see Configuration section). It searches the content with


FAQ

What is the Probe Documentation Search MCP server?
Probe Documentation Search is a Model Context Protocol (MCP) server profile on explainx.ai. MCP lets AI hosts (e.g. Claude Desktop, Cursor) call tools and resources through a standard interface; this page summarizes categories, install hints, and community ratings.
How do MCP servers relate to agent skills?
Skills are reusable instruction packages (often SKILL.md); MCP servers expose live capabilities. Teams frequently combine both—skills for workflows, MCP for APIs and data. See explainx.ai/skills and explainx.ai/mcp-servers for parallel directories.
How are reviews shown for Probe Documentation Search?
This profile displays 10 aggregated ratings (sample rows for discoverability plus signed-in user reviews). Average score is about 4.5 out of 5—verify behavior in your own environment before production use.
MCP server reviews

Ratings

4.510 reviews
  • Shikha Mishra· Oct 10, 2024

    Probe Documentation Search is among the better-indexed MCP projects we tried; the explainx.ai summary tracks the official description.

  • Piyush G· Sep 9, 2024

    We evaluated Probe Documentation Search against two servers with overlapping tools; this profile had the clearer scope statement.

  • Chaitanya Patil· Aug 8, 2024

    Useful MCP listing: Probe Documentation Search is the kind of server we cite when onboarding engineers to host + tool permissions.

  • Sakshi Patil· Jul 7, 2024

    Probe Documentation Search reduced integration guesswork — categories and install configs on the listing matched the upstream repo.

  • Ganesh Mohane· Jun 6, 2024

    I recommend Probe Documentation Search for teams standardizing on MCP; the explainx.ai page compares cleanly with sibling servers.

  • Oshnikdeep· May 5, 2024

    Strong directory entry: Probe Documentation Search surfaces stars and publisher context so we could sanity-check maintenance before adopting.

  • Dhruvi Jain· Apr 4, 2024

    Probe Documentation Search has been reliable for tool-calling workflows; the MCP profile page is a good permalink for internal docs.

  • Rahul Santra· Mar 3, 2024

    According to our notes, Probe Documentation Search benefits from clear Model Context Protocol framing — fewer ambiguous “AI plugin” claims.

  • Pratham Ware· Feb 2, 2024

    We wired Probe Documentation Search into a staging workspace; the listing’s GitHub and npm pointers saved time versus hunting across READMEs.

  • Yash Thakker· Jan 1, 2024

    Probe Documentation Search is a well-scoped MCP server in the explainx.ai directory — install snippets and categories matched our Claude Code setup.