G2 MCP Server

The G2 MCP Server enables you to connect AI assistants to G2 buyer intent data, competitive intelligence, product catalog data, and review analytics.

Use this server to identify companies researching your products, understand which competitors they evaluate, prioritize accounts for outreach, and analyze review data through natural language prompts.

Basics of G2 MCP Server

Features

• Buyer Intent Intelligence: Track companies researching your products, review intent scores, and identify accounts for prioritization.
• Competitive Intelligence: Discover which competitors prospects evaluate and review comparison patterns.
• Review Analytics: Access review data with market intelligence context including company segments, industries, feature ratings, and competitive switching patterns

MCP-enabled partners

The G2 MCP Server can be used with supported partner platforms to bring G2 data into AI agents and workflow automation tools.

HubSpot

The G2 MCP for HubSpot enables Breeze Agents to access G2 buyer intent data, review data, and competitor intelligence in real time.

Breeze Agents can access buyer intent signals, competitor intelligence, and review data to support account prioritization, pre-call preparation, and account research.

The following datasets are available through the G2 MCP Server:

• Buyer Intent Signals
• Product Reviews
• Product Catalog
• Category Taxonomy

Requirements

• HubSpot subscription with access to Breeze Agents
• G2 subscription
• Additional G2 entitlements may be required for buyer intent data

Limitations

• Read-only access to G2 data
• Available through Breeze Agent workflows only
• Does not write data to HubSpot records by default

Gong

The G2 MCP for Gong enables Gong AI workflows to access G2 buyer intent data, competitive intelligence, and review insights through natural language prompts.

Gong users can prioritize accounts using buyer intent signals, generate pre-call briefs, identify competitors in active deals, and surface buyer research activity directly within Gong workflows. G2 MCP provides conversational access to G2 data through AI-driven workflows.

Available data

• Buyer Intent Signals
• Competitive Signals
• Buying Stage and Activity Level
• Review Insights and Customer Sentiment
• Category and Comparison Research Activity

Requirements

• G2 subscription
• Gong subscription
• Access to Buyer Intent data and Gong AI workflows

Limitations

• Available through Gong AI workflows only
• Access to G2 data depends on your G2 subscription and entitlements

AirOps

The G2 MCP for AirOps enables AirOps Playbooks to access G2 review intelligence and market data.

Teams can use G2 data to create review-informed landing pages, competitor comparison content, SEO and AI search optimized articles, social proof assets, and competitive battlecards.

Available data

• Product Reviews

• Product Catalog

• Category Taxonomy

• Competitive Signals and Comparison Activity

• Buying Stage and Category Research Behavior*

• G2 Quarterly Grid Reports (coming soon)

• Available for customers licensed for Buyer Intent.

Requirements

• AirOps subscription
• G2 subscription
• Access to AirOps Playbooks

Limitations

• Available through AirOps Playbooks only
• Access to G2 data depends on your G2 subscription and entitlements
• Grid Report data is not yet available

Profound

The G2 MCP for Profound enables Profound Agents and Ask Profound to access G2 buyer intent data, review intelligence, and market research data through natural language prompts.

Marketing teams can use G2 data within Profound to understand how AI systems perceive their brand, analyze customer sentiment, identify competitive positioning, and inform AI search optimization strategies using verified customer voice and buyer research activity.

Available data

• Buyer Intent Signals

• Product Reviews

• Product Catalog

• Category Taxonomy

• Competitive Signals and Comparison Activity

• Buying Stage and Category Research Behavior*

• Available for customers licensed for Buyer Intent.

Requirements

• Profound subscription
• G2 subscription
• Access to Profound Agents or Ask Profound

Limitations

• Available through Profound Agents and Ask Profound only
• Access to G2 data depends on your G2 subscription and entitlements
• Read-only access to G2 data

Accessing G2 MCP

Choose the connection method that best fits your workflow:

• Claude and ChatGPT provide a no-code experience for interacting with G2 data using natural language.
• Direct MCP server connections are intended for developers and partners building custom AI agents or integrating G2 into their own applications.

All connection methods use OAuth 2.0 to authenticate with your G2 account.

Before you begin, you'll need:

• A valid G2 account with appropriate access permissions.
• Products registered on G2 to access the features described in this guide.
• The required G2 subscription and entitlements for the data and features you want to access.

Claude

Use the following steps to connect the G2 MCP connector in Claude.

1. In Claude Desktop or Claude on the web, select the + icon below the chat box.
2. Select Connectors > Manage Connectors.
3. Search for G2.
4. Select G2 MCP.
5. Select Connect.
6. In the browser window that opens, select Sign Up.
7. Complete sign-up, and then return to Claude.
8. Start a new chat.
9. Select the + icon again and confirm that G2 MCP is turned on.

ChatGPT

Use the following steps to connect the G2 MCP connector in ChatGPT.

1. Open ChatGPT, and then select Apps from the left navigation.
2. Search for and select G2 MCP.
3. In the Connect to G2 MCP window, complete the OAuth flow by signing in to your G2 account.
4. Return to New chat.
5. Select the + icon, then select G2 MCP.
6. Verify that G2 MCP appears in blue in the chat window. This indicates that the connector is active.

Connect directly to the G2 MCP server

You can connect directly to the G2 MCP server if you're building a custom Model Context Protocol (MCP) client or integrating G2 into your own agentic workflow.

This connection method uses OAuth 2.0 to authenticate with your G2 account and supports any MCP-compatible client that can connect to a remote MCP server.

Prerequisites

Before you connect to the G2 MCP server, you need:

• A registered G2 OAuth application with:
  • A client_id
  • A client_secret
  • http://localhost:<PORT>/callback added as an allowed redirect URI

To register or manage your OAuth application, visit the G2 Developer Dashboard.

Request only the OAuth scopes your application requires.

Authentication

The G2 MCP server uses OAuth 2.0 to authenticate requests. During authentication:

1. Your client discovers the authorization server.
2. Your client opens a browser and starts the OAuth Authorization Code with PKCE flow.
3. After you authorize the application, G2 redirects to your local callback URL with an authorization code.
4. Your client exchanges the authorization code for an access token.
5. Your client sends the access token as a Bearer token with requests to the G2 MCP server.
6. The G2 MCP server validates the access token before processing requests.

OAuth tokens are cached for the current session and refreshed automatically.

Python

The following example uses FastMCP, which manages the OAuth browser flow automatically.

To use the Python example, install:

• Python 3.12 or later
• uv

Install FastMCP:

uv add fastmcp

Example

import asyncio
import os
from fastmcp import Client
from fastmcp.client.auth import OAuth

SERVER_URL = os.getenv("G2_MCP_URL", "https://mcp.g2.com/mcp")
CLIENT_ID = os.getenv("G2_CLIENT_ID", "")
CLIENT_SECRET = os.getenv("G2_CLIENT_SECRET", "")

CALLBACK_PORT = 8080

SCOPES = [
    "openid",
    "profile",
    "buyer_intent.read",
    "products.read",
    "products.reviews.read",
    "vendors.read",
]

async def main() -> None:
    if not CLIENT_ID or not CLIENT_SECRET:
        raise RuntimeError("Set G2_CLIENT_ID and G2_CLIENT_SECRET.")

    auth = OAuth(
        client_id=CLIENT_ID,
        client_secret=CLIENT_SECRET,
        scopes=SCOPES,
        callback_port=CALLBACK_PORT,
    )

    async with Client(SERVER_URL, auth=auth) as client:
        await client.ping()

        tools = await client.list_tools()
        for tool in tools:
            print(tool.name)

        result = await client.call_tool("list_products", {"page_size": 3})
        print(result.data)

if __name__ == "__main__":
    asyncio.run(main())

A complete example is available in client/example_client.py.

Run the client:

G2_CLIENT_ID=your_client_id \
G2_CLIENT_SECRET=your_client_secret \
uv run python client/example_client.py

If you've already configured the environment variables, run:

uv run python client/example_client.py

The first time you run the client, a browser window opens so you can authorize the application with your G2 account. After authentication, the client connects to the G2 MCP server, lists the available tools, and calls list_products.

TypeScript

The following example uses the official MCP TypeScript SDK.

To use the TypeScript example, install:

• Node.js 18 or later
• npm

Install the required packages:

npm install @modelcontextprotocol/client @cfworker/json-schema open

The example also requires the simpleOAuthClientProvider.ts helper included with the MCP SDK examples. The file is located at:

client/typescript/simpleOAuthClientProvider.ts

Example

The complete implementation is available in client/typescript/g2_mcp_client.ts.

Configure your OAuth provider using your pre-registered client credentials:

const CLIENT_ID = process.env.G2_CLIENT_ID ?? '';
const CLIENT_SECRET = process.env.G2_CLIENT_SECRET ?? '';

const CALLBACK_URL = 'http://localhost:8080/callback';

Run the client:

G2_CLIENT_ID=your_client_id \
G2_CLIENT_SECRET=your_client_secret \
npm start

After authentication, you can interact with the server using commands such as:

mcp> list
mcp> call list_products {"page_size": 3}
mcp> call show_product {"product_id":"<uuid>"}
mcp> stream browse_buyer_intent {"subject_product_id":"<uuid>"}
mcp> quit

G2 OAuth configuration

The G2 OAuth implementation differs from the default MCP TypeScript SDK examples. Update the example as described in the following sections to connect successfully to the G2 MCP server.

Use pre-registered client credentials

The default MCP TypeScript SDK uses Dynamic Client Registration (DCR) to register a client automatically on first run. G2's OAuth server doesn't support DCR.

Configure InMemoryOAuthClientProvider with your pre-registered client_id, client_secret, and redirect URI instead.

const oauthProvider = new InMemoryOAuthClientProvider(
    CALLBACK_URL,
    clientMetadata,
    (redirectUrl: URL) => {
        this.openBrowser(redirectUrl.toString());
    },
    this.clientMetadataUrl,
    {
        client_id: CLIENT_ID,
        client_secret: CLIENT_SECRET,
        redirect_uris: [CALLBACK_URL],
    },
);

Providing these credentials bypasses Dynamic Client Registration and uses your existing OAuth application.

Handle the OAuth callback

The default MCP TypeScript SDK example expects the OAuth callback to include both a code and state parameter.

G2's OAuth server returns the authorization code but doesn't include the state parameter. Update your callback handler to require only the authorization code.

Token introspection requirements

The G2 MCP server validates access tokens by calling G2's token introspection endpoint using its own server-side credentials.

For token validation to succeed, one of the following must be true:

• The MCP server's OAuth application has cross-application token introspection enabled.
• Your client uses the same client_id configured for the MCP server.

Supported MCP tools

After connecting to the G2 MCP server, you can discover available tools by calling list_tools().

For parameter definitions and request details, see the G2 API documentation.

Usage Examples

Example 1: Researching companies for sales outreach

User prompt: "Show me companies researching G2 Ads with their intent scores and which competitors they're comparing."

What happens:

• Server retrieves buyer intent signals for companies interested in your product.
• Returns company names, intent scores (0-100 scale), and activity levels.
• Identifies competitor products being evaluated (e.g., StackAdapt, The Trade Desk, Adobe Advertising Cloud).
• Provides firmographic context (company size, location, domain).

Example 2: Prioritizing accounts for prospecting

Focus outreach on accounts with active buying signals
User prompt: "Which accounts should I prioritize this week based on G2 intent signals?"

What happens:

• Server queries recent buyer intent activity across your products.
• Ranks accounts by intent score and engagement level (activity count, visitor count).
• Returns prioritized list with company details (name, domain, size, location).
• Highlights enterprise opportunities and high-intent prospects for immediate follow-up.

Example 3: Preparing a pre-call briefing

User prompt: "Give me a pre-call briefing for Confluent using G2 data."

What happens:

• Server retrieves recent buyer intent activity for the specified company.
• Identifies specific competitors the company is researching (e.g., The Trade Desk, Adobe).
• Summarizes signal types (profile views, comparison pages, pricing views).
• Provides day-level activity timeline to show buying urgency.
• Returns actionable insights: intent score, competitive context, and recommended talking points.

Privacy Policy

See G2's Privacy Policy.

Support

• G2 Support
• G2 Documentation
• G2 V2 API Documentation
• General G2 Help Center