Argorant
Argorant MCP Server — give your AI agent direct access to 614M verified B2B contacts. Query by industry, role, geography, and 100+ filters, then export emails verified by a live SMTP probe at request time (catch-alls flagged, invalids free). OAuth-secured. Works with Claude, ChatGPT, Cursor, and any MCP client. Endpoint: https://mcp.argorant.com/mcp
README
Argorant Public MCP
Hosted remote MCP server for Argorant.
- MCP endpoint:
https://mcp.argorant.com/mcp - Documentation:
https://argorant.com/docs/mcp/overview - Setup guide:
https://argorant.com/docs/mcp/connect - Tool reference:
https://argorant.com/docs/mcp/tools - ChatGPT setup:
https://argorant.com/docs/mcp/openai - Claude setup:
https://argorant.com/docs/mcp/claude - Security boundaries:
https://argorant.com/docs/mcp/security - Submission checklist:
https://argorant.com/docs/mcp/submission
This is a hosted remote MCP connector. Claude, ChatGPT, and custom clients that support remote MCP should connect to the endpoint directly; no npm package is required for that flow. A local package can still be added later as an optional bridge, but it should not be required for the public hosted connector.
The public server uses OAuth and account-scoped access. The first tools are read-only by design:
searchfetchargorant_accountargorant_count_peopleargorant_preview_people
search and fetch support ChatGPT apps, deep
research, and other clients that expect data-only search/fetch semantics. They
return aggregate coverage summaries and capped masked preview rows only.
Contact reveal, phone reveal, profile URLs, list creation, and exports are disabled by default. Scoped reveal/export tools can only run after OAuth, workspace permission, quota, plan, and audit checks. Use the signed-in Argorant workspace for normal review and download workflows.
Directory Readiness
Use the hosted endpoint for public review:
https://mcp.argorant.com/mcp
Before submitting to ChatGPT, Claude, or an MCP directory, verify:
- The endpoint is reachable over public HTTPS.
- OAuth protected-resource and authorization-server metadata resolve on the MCP host.
- The connector manifest includes setup, security, privacy, terms, and submission links.
- Tools are read-only and return only counts, safe safe result links, or masked previews.
- Raw emails, phone numbers, profile URLs, reveal actions, exports, write actions, and destructive actions remain disabled by default.
- Reviewers have a normal member account and an owner/admin account for usage-limit testing.
Discovery
Clients should discover OAuth metadata from the MCP host:
https://mcp.argorant.com/.well-known/oauth-protected-resource
https://mcp.argorant.com/.well-known/oauth-protected-resource/mcp
https://mcp.argorant.com/.well-known/oauth-authorization-server
Unauthenticated requests to /mcp return a 401 response with a
WWW-Authenticate header that points to protected-resource metadata.
Argorant also publishes a connector manifest for client directories and setup flows:
https://mcp.argorant.com/manifest.json
https://mcp.argorant.com/.well-known/argorant-mcp.json
The manifest describes the MCP endpoint, docs, OAuth scopes, safe tools, installation mode, and data-exposure defaults. It does not grant access by itself.
Local Service
The production service runs under systemd:
systemctl status argorant-public-mcp.service
systemctl restart argorant-public-mcp.service
The ASGI entrypoint is:
argorant_mcp.app:app
Default runtime settings come from environment variables in systemd:
MCP_PUBLIC_BASE_URLMCP_PATHARGORANT_BACKEND_BASE_URLARGORANT_BACKEND_PUBLIC_URLARGORANT_MCP_BRIDGE_SECRETMCP_DATABASE_URL
Smoke Checks
curl -sS https://mcp.argorant.com/ | python3 -m json.tool
curl -sS https://mcp.argorant.com/manifest.json | python3 -m json.tool
curl -sS https://mcp.argorant.com/.well-known/argorant-mcp.json | python3 -m json.tool
curl -sSI https://mcp.argorant.com/mcp
curl -sS https://mcp.argorant.com/.well-known/oauth-protected-resource | python3 -m json.tool
curl -sS https://mcp.argorant.com/.well-known/oauth-protected-resource/mcp | python3 -m json.tool
curl -sS https://mcp.argorant.com/.well-known/oauth-authorization-server | python3 -m json.tool
Expected behavior:
/returns service metadata and links to docs./manifest.jsonand/.well-known/argorant-mcp.jsonreturn connector metadata./mcpreturns401without a bearer token.- OAuth discovery endpoints return JSON metadata.
- App, MCP, and API hosts stay separated:
argorant.comfor marketing, docs, tools, and research pages.app.argorant.comfor login and the signed-in app.mcp.argorant.comfor remote MCP and OAuth metadata.
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。