muro MCP server
The official hosted MCP server for muro.chat, enabling AI agents to manage customer support conversations, send replies, triage tickets, and configure the live chat widget.
README
muro MCP server
The official hosted Model Context Protocol server for muro.chat, the live chat widget + AI support inbox for people running multiple projects.
Connect your AI agent (Claude, Claude Desktop, Cursor, Windsurf, ChatGPT, or any MCP client) to your support inbox: it can list and search customer conversations, read full message threads, send replies or internal notes, triage with status/tags/assignee, look up contacts, and even configure the chat widget. Support becomes a tool call inside your editor.
There is nothing to install and no local process: the server is a remote Streamable HTTP endpoint.
https://muro.chat/api/mcp
Also listed on the official MCP registry as chat.muro/support.
Authentication
Create an API key in the muro dashboard (Settings → API keys). Keys look like mr_live_… and carry scopes:
| Scope | Allows |
|---|---|
read |
list/search conversations, read threads, contacts, sites |
write |
send replies and notes, triage, update widget config |
A key can be pinned to a single site: every request it makes (REST and MCP alike) only sees that site's conversations, contacts and config. One key per repo means the agent working on a project only ever sees that project's inbox.
Send the key as a Bearer token:
Authorization: Bearer mr_live_…
Quick start
Clients that support remote MCP servers natively can use the URL + header directly. stdio-only clients connect through mcp-remote:
npx mcp-remote https://muro.chat/api/mcp \
--header "Authorization: Bearer mr_live_…"
This repo also ships a tiny open-source bridge, index.js, that wraps mcp-remote with the muro endpoint baked in; it reads the key from the MURO_API_KEY environment variable. Run it from a checkout (npm install && node index.js) or via Docker:
docker build -t muro-mcp .
docker run -i -e MURO_API_KEY=mr_live_... muro-mcp
Ready-to-paste configs for common clients are in examples/:
- Claude Desktop (
claude_desktop_config.json) - Cursor (
.cursor/mcp.json) - Windsurf (
~/.codeium/windsurf/mcp_config.json)
Tools
| Tool | What it does |
|---|---|
list_conversations |
List conversations; filter by status, full-text search |
get_conversation |
One conversation with its messages |
send_reply |
Reply to the visitor, or leave an internal note |
update_conversation |
Change status, tags, assignee |
list_contacts |
List contacts |
get_contact |
One contact's details |
list_sites |
Your sites and their widget config |
update_site |
Configure the widget: color, position, welcome message, language, size |
Example prompts
Once connected, ask your agent things like:
- "What are customers complaining about today?"
- "Read the last conversation about the checkout bug and summarize it."
- "Draft a reply to the visitor asking about refunds, friendly tone."
- "Close every conversation we already answered this week."
- "Switch the widget on my landing page to dark blue and French."
Links
- Product: muro.chat
- API + MCP docs: muro.chat/docs/api
- Live demo: muro.chat/demo
- Contact: hello@muro.chat
License
The documentation and example configurations in this repository are MIT licensed (see LICENSE). The muro service itself is a commercial product.
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。