Potarix enricher
Find contact/websites for any company/person
README
Potarix MCP Server
MCP wrapper for Potarix Enricher. Lets AI agents resolve company websites, find verified emails, and pull complete company rosters — and (with one human-in-the-loop card capture) sign up and pay for credits entirely from the agent.
<!-- mcp-name: io.github.Potarix/potarix-mcp -->
Tools
| tool | what it does | cost |
|---|---|---|
lookup_company_website |
company name → website URL | 2 credits |
find_person_email |
named person + company/domain → verified email | 25 credits |
find_decision_maker_email |
category + domain → likely buyer name + email | 25 credits |
find_linkedin_email |
LinkedIn profile URL → verified email | 10 credits |
find_company_emails |
domain → public company contact roster | 25 credits |
find_all |
one company name → website + DMs + full company email list | sum of above |
check_balance |
credits, email, saved-card status, key count | free |
start_checkout |
get a Stripe URL to add a card the first time | n/a |
topup_credits |
charge the saved card and add credits | n/a |
1 credit = $0.01. Trial accounts start with 25 free credits. Every endpoint floors at the worst-case provider COGS — a hit never loses money, and short-circuited waterfall calls earn margin.
Two ways to connect
| transport | endpoint | best for |
|---|---|---|
| Streamable HTTP (hosted) | https://api.potarix.com/mcp |
remote agents (Claude, ChatGPT) — no install |
| stdio (npm package) | npx -y potarix-mcp |
local/desktop clients |
Both expose the same nine tools. The hosted server is stateless and multi-tenant: every request is authorized by its own Authorization: Bearer ptk_live_... header, so there is no per-user deployment.
Hosted (Streamable HTTP)
Point any MCP client that speaks Streamable HTTP at https://api.potarix.com/mcp and send your key as a bearer token:
{
"mcpServers": {
"potarix": {
"url": "https://api.potarix.com/mcp",
"headers": { "Authorization": "Bearer ptk_live_your_key" }
}
}
}
Install (stdio)
npm install -g potarix-mcp
Or run it without a global install:
npx -y potarix-mcp
Configure
Set your Potarix API key:
export POTARIX_API_KEY=ptk_live_your_key
Optional:
export POTARIX_API=https://api.potarix.com/enricher
Claude Desktop
{
"mcpServers": {
"potarix": {
"command": "npx",
"args": ["-y", "potarix-mcp"],
"env": {
"POTARIX_API_KEY": "ptk_live_your_key"
}
}
}
}
Claude Code
claude mcp add potarix npx -- -y potarix-mcp
Then add POTARIX_API_KEY to the environment where Claude Code runs.
Development
npm install
npm run build
npm run smoke # stdio transport: connect + tools/list
# Streamable HTTP transport:
npm run start:http # serves on http://127.0.0.1:8080/mcp (set PORT to change)
# in another shell:
POTARIX_MCP_URL=http://127.0.0.1:8080/mcp npm run smoke:http
Environment variables:
| var | default | purpose |
|---|---|---|
POTARIX_API_KEY |
— | API key (stdio only; HTTP reads it per-request from the bearer header) |
POTARIX_MCP_TRANSPORT |
stdio |
http to run the Streamable HTTP server (or pass --http) |
PORT / HOST |
8080 / 127.0.0.1 |
HTTP bind address |
POTARIX_MCP_PATH |
/mcp |
HTTP request path |
POTARIX_MCP_ALLOWED_HOSTS |
— | comma-separated host allow-list; enables DNS-rebinding protection |
POTARIX_API |
https://api.potarix.com/enricher |
API base URL override |
Registry Publishing
This repo includes server.json for the official MCP Registry. The entry is
multi-surface: it declares both the hosted Streamable HTTP remote
(https://api.potarix.com/mcp) and the npm stdio package (potarix-mcp), so a
single registry record advertises two ways to connect.
The registry validates the npm package by fetching it and checking that its
published mcpName matches the server name, so the npm package must be
published first, at the same version named in server.json (packages[].version).
Publishing steps (version-bump first, then npm, then registry):
# 1. Bump package.json + server.json to the same new version (e.g. 0.1.3).
# server.json must be a NEW version each publish (versions are immutable).
# 2. Build + publish the npm artifact (carries mcpName for ownership proof):
npm publish
# 3. Push the multi-surface server.json to the official MCP Registry:
mcp-publisher login github
mcp-publisher publish
The package mcpName in package.json must match server.json:
io.github.Potarix/potarix-mcp
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。