Gong MCP
a full-surface Model Context Protocol gateway to the Gong REST API (calls, transcripts, users, activity, scorecards, coaching, CRM). Full API coverage, zero prerequisites, bring your own credentials.
README
<p align="center"> <img src="assets/social-preview.png" alt="Gong MCP — the full-surface Gong API gateway for Claude" width="820"> </p>
<h1 align="center">Gong MCP</h1>
<p align="center"> <b>The full-surface <a href="https://modelcontextprotocol.io">Model Context Protocol</a> server for the <a href="https://www.gong.io">Gong</a> API.</b><br> 20 tools across calls, transcripts, users, activity, scorecards, coaching and CRM. Install into Claude Desktop in one click. </p>
<p align="center"> <a href="https://github.com/scandolo/gong-mcp/actions/workflows/ci.yml"><img src="https://github.com/scandolo/gong-mcp/actions/workflows/ci.yml/badge.svg" alt="CI"></a> <img src="https://img.shields.io/badge/tools-20-7c5cff" alt="20 tools"> <img src="https://img.shields.io/badge/MCP-server-6a4bff" alt="MCP server"> <a href="./LICENSE"><img src="https://img.shields.io/github/license/scandolo/gong-mcp" alt="MIT license"></a> </p>
⬇️ Install in Claude Desktop (one click, no terminal)
If you use the Claude Desktop app, this is the easy path. No Node, no JSON, no command line.
<p> <a href="https://github.com/scandolo/gong-mcp/releases/latest/download/gong-mcp.mcpb"> <img src="https://img.shields.io/badge/⬇%20Download%20the%20Gong%20MCP%20extension-.mcpb-7c5cff?style=for-the-badge" alt="Download the Gong MCP extension"> </a> </p>
- Download
gong-mcp.mcpb(one file). - Double-click it (or in Claude Desktop go to Settings → Extensions → Install extension… and pick the file).
- When prompted, paste your Gong base URL, Access Key, and Access Secret (see how to get them), then click Install.
That is it. Ask Claude "list my Gong workspaces" to confirm it is connected.
<p align="center"> <img src="assets/demo.gif" alt="Claude using Gong MCP to pull a call transcript and summarise renewal risk" width="720"> </p>
Prefer Claude Code / Cursor / a terminal? Jump to Install in Claude Code.
Gong MCP vs the official Gong MCP
Gong ships an official hosted MCP server at mcp.gong.io. It is a different tool for a different job: three AI tools that return generated answers and briefs. This repo exposes the raw Gong REST API instead, so you get the underlying data and can build your own analysis on top.
Official Gong MCP (mcp.gong.io) |
This repo (gong-mcp) |
|
|---|---|---|
| Tools | 3 (ask_account, ask_deal, generate_brief) |
20 (raw REST API surface) |
| Returns | AI-generated answers and briefs | raw data: transcripts, calls, users, activity, scorecards, coaching, CRM |
| Raw transcripts / call lists | not returned | yes |
| Grain | account and deal | call, user, workspace, CRM |
| Hosting | Gong-hosted remote connector | self-hosted, runs locally |
| Auth and setup | admin creates an MCP integration, MCP scope, Client ID/Secret | your own Gong API key and secret |
| Cost | consumes Gong credits | free (your normal API quota) |
| Claude plan | Pro/Max/Team/Enterprise (custom remote connectors) | any (local stdio server) |
Rule of thumb: use the official server for quick AI briefs with zero infrastructure; use this one when you need raw transcripts, call and activity data, or want to build your own analysis over the full API.
Tools
| Tool | What it does |
|---|---|
list_workspaces |
List Gong workspaces / confirm auth. |
list_calls |
List calls in a date range (light metadata). |
get_call |
Get one call's metadata by id. |
list_calls_extensive |
Rich call data: content, parties, stats, CRM context. |
get_transcripts |
Word-level call transcripts. |
list_library_folders |
List library folders for a workspace. |
get_library_folder_content |
List calls in a library folder. |
list_users |
List all Gong users. |
get_user |
Get one user by Gong id. |
get_users_extensive |
Batch user lookup by id or email. |
get_activity_aggregate |
Per-user activity totals. |
get_activity_by_period |
Per-user activity by period. |
get_activity_day_by_day |
Per-user, per-day activity. |
get_interaction_stats |
Rep interaction quality stats. |
list_scorecards |
List scorecard templates. |
get_answered_scorecards |
Actual answered scorecards. |
get_coaching_metrics |
Manager coaching metrics. |
get_manual_crm_associations |
Manual call-to-CRM associations. |
list_crm_integrations |
List CRM integrations. |
get_crm_entities |
Read custom-integration CRM objects. |
Getting your Gong credentials
- In Gong, go to Company Settings → API.
- Click Get API Key (Technical Administrator role required).
- Copy the Access Key and Access Key Secret.
- Note your workspace host (region-specific), which looks like
https://us-XXXXX.api.gong.io. It is shown alongside your API key in Gong. If left blank, the server falls back tohttps://api.gong.io.
Credentials are read from environment variables (
GONG_ACCESS_KEY,GONG_ACCESS_SECRET,GONG_BASE_URL) or the Claude Desktop install dialog. Nothing is hardcoded and nothing is committed to this repo.
Install in Claude Code
For the CLI / terminal agent (and Cursor, or any MCP client).
git clone https://github.com/scandolo/gong-mcp.git
cd gong-mcp
npm install
# register the server with Claude Code (user scope)
claude mcp add gong node "$(pwd)/server/index.js" \
--scope user \
--env GONG_BASE_URL=https://us-XXXXX.api.gong.io \
--env GONG_ACCESS_KEY=your-access-key \
--env GONG_ACCESS_SECRET=your-access-secret
Or add it directly to ~/.claude.json under mcpServers:
{
"mcpServers": {
"gong": {
"command": "node",
"args": ["/absolute/path/to/gong-mcp/server/index.js"],
"env": {
"GONG_BASE_URL": "https://us-XXXXX.api.gong.io",
"GONG_ACCESS_KEY": "your-access-key",
"GONG_ACCESS_SECRET": "your-access-secret"
}
}
}
}
Restart Claude Code and ask it to run list_workspaces to confirm auth.
Copy-paste prompt (Claude Code): paste this into a Claude Code session, filling in your three values.
Set up the Gong MCP server (https://github.com/scandolo/gong-mcp) in this Claude Code
environment for me. Do the following:
1. Clone https://github.com/scandolo/gong-mcp into ~/code/gong-mcp (or tell me if it
already exists), then run `npm install` in it.
2. Register it with Claude Code at user scope using `claude mcp add`, with these env vars:
GONG_BASE_URL=<PASTE YOUR HOST, e.g. https://us-XXXXX.api.gong.io>
GONG_ACCESS_KEY=<PASTE YOUR GONG ACCESS KEY>
GONG_ACCESS_SECRET=<PASTE YOUR GONG ACCESS SECRET>
3. Verify the server is registered with `claude mcp list`.
4. Tell me to restart Claude Code, then confirm it works by calling the `list_workspaces` tool.
Never print my access key or secret back to me or write them into any tracked/committed file.
Build the Claude Desktop bundle yourself
The prebuilt .mcpb is attached to every release. To build it from source:
npm install
npx @anthropic-ai/mcpb pack # produces gong-mcp.mcpb
Configuration
| Env var | Required | Default | Notes |
|---|---|---|---|
GONG_ACCESS_KEY |
yes | — | Gong API Access Key. |
GONG_ACCESS_SECRET |
yes | — | Gong API Access Key Secret. |
GONG_BASE_URL |
no | https://api.gong.io |
Your workspace-specific host, e.g. https://us-XXXXX.api.gong.io. |
Authentication uses HTTP Basic auth (AccessKey:Secret, base64-encoded) exactly as the Gong REST API expects. The client retries on transient errors and rate limits with exponential backoff.
Requirements
- Node.js >= 18 (bundled with Claude Desktop; install separately for Claude Code)
- A Gong seat with API access (Technical Administrator to mint keys)
Contributing
Issues and PRs welcome, especially additional Gong endpoints as the API grows. See the issue templates when opening a report.
License
<sub>Keywords: Gong MCP, Gong MCP server, Gong API MCP, Gong Model Context Protocol, Gong API gateway, Gong Claude integration, Gong Claude Desktop extension, Gong conversation intelligence MCP, Gong REST API wrapper, RevOps MCP, sales intelligence MCP.</sub>
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。