> Enrich.chat gives AI agents verified work emails and mobiles. To use it from an agent, read https://enrich.chat/agents.md . This is the Markdown version of https://enrich.chat/features/mcp-server/

# MCP server for contact data: use Enrich.chat in Claude, ChatGPT and Cursor

Add one URL to your AI app and it can find people, work emails, mobile numbers and companies. Try it with no sign-in, or sign in with your account over OAuth to use your plan.

By Enrich.chat Editorial. Updated October 19, 2026.

**Short answer:** Enrich.chat runs a remote MCP server with 11 tools that find people, work emails, mobile numbers and company data inside your AI app. Add `enrich.chat/mcp` to try it with no sign-in and 10 free lookups a day, or `enrich.chat/mcp/oauth` to sign in with your account over OAuth and use your plan's credits.

## Key takeaways

- The server has 11 tools at two addresses: `enrich.chat/mcp` (no sign-in, 10 free lookups a day) and `enrich.chat/mcp/oauth` (your account and your plan's credits).
- The OAuth address uses OAuth 2.1 with PKCE and dynamic client registration. You log in with an emailed code and press Allow, and the app cannot see your account details, API keys or billing.
- Credits match the API: 10 per work email, 250 per mobile, 5 per profile or per person returned in a search, and 0 when nothing is found.
- Disconnect any app in the dashboard under Connected apps on the Connect an agent screen. It stops working at once.
- The search tools take the extra filters through `more_filters`, an object of filter names and values, and `count_only` gives the free count.

## Contact data inside the AI app you already use

- **One URL, 11 tools.** People, work emails, mobiles, companies, tech stacks and email formats, called from a normal chat.
- **Try it with no sign-in.** enrich.chat/mcp gives 10 free lookups a day with no signup. A lookup that finds nothing is free.
- **Sign in with OAuth.** enrich.chat/mcp/oauth uses OAuth 2.1 with PKCE. Log in, press Allow, and the app uses your plan.
- **Read-only tools.** Every tool is annotated read-only, and your app can ask before each call.
- **Works where you work.** Claude, ChatGPT, Cursor, VS Code, Claude Code and Codex all take a remote server URL.
- **Credits on matches only.** 10 credits per work email, 250 per mobile, 5 per profile. Nothing found uses 0.

## From a URL to a contact in three steps

1. **Add the URL.** Paste `enrich.chat/mcp` or `enrich.chat/mcp/oauth` into your app's connector or MCP settings.
2. **Sign in if asked.** On the OAuth address, log in with your email and a 6-digit code, then press Allow.
3. **Ask in plain English.** Say who you want. The app picks the tool and shows what it found, with the status of each field.

## What does the Enrich.chat MCP server do?

The Enrich.chat MCP server lets an AI app look up people, companies and contact details by calling tools. MCP, the Model Context Protocol, is [an open standard for connecting AI applications to external systems](https://modelcontextprotocol.io/docs/getting-started/intro). Enrich.chat runs a remote server, so there is nothing to install. You paste one URL into your app.

The server has 11 tools. Each one calls an endpoint of the [Enrich.chat API](https://enrich.chat/features/api/), so credits, limits and opt-outs work the same way as they do over HTTP. Each tool is annotated as read-only, so none of them changes anything in your accounts. A lookup that finds a contact still uses credits. The MCP specification says [there should always be a human able to deny a tool call](https://modelcontextprotocol.io/specification/2026-07-28/server/tools), so your app may ask before it runs one.

| Tool | What it does | Credits when found |
| --- | --- | --- |
| `find_person_contact` | One person's work email, mobile and profile, from a LinkedIn URL, person id, work email, or name plus company website | 10 per email, 250 per mobile, 5 per profile |
| `find_work_email` | A work email from a name and company website, a LinkedIn URL or a person id | 10 |
| `find_mobile_number` | A mobile number from a LinkedIn URL, person id or work email | 250 |
| `search_people` | People by plain English or filters, with extra filters in `more_filters`. `count_only` is free | 5 per person returned |
| `company_people` | People at one company website, by title or level | 5 per person returned |
| `search_companies` | Companies by industry, place, size, revenue, technology, funding and growth, with extra filters in `more_filters`. `count_only` is free | 5 per company returned |
| `company_profile` | One company's profile, with technologies on request | 5 |
| `company_tech_stack` | The technologies a company website uses | 5 |
| `company_email_format` | The email formats a company uses, and how common each is | 5 |
| `list_filter_values` | The exact values a filter accepts | 0 |
| `check_free_quota` | How many free lookups are left today | 0 |

The two search tools take the same kinds of filters as the dashboard's [people search](https://enrich.chat/features/people-search/) and [company search](https://enrich.chat/features/company-search/). Beyond the common ones, pass the extra filters in `more_filters`, an object of filter names and values such as `{"skills": "python"}`. `list_filter_values` shows the exact values a filter accepts, and the API's `GET /v1/filters` lists every filter with an example.

### What does a tool call cost for a real task?

Say you ask Claude for the CFO of a company and whether their work email is verified. The app calls `company_people` with the company website and a title, then `find_person_contact` with the `person_id` from the answer. The code tabs at the top of this page show the first call and a result. The people and companies in them are made up.

For a bigger pass, 100 people from a search use 500 credits, and 100 work emails add 1,000. That is 1,500 credits, which fits the Growth plan at $29 a month. On the free tier, one search uses one lookup however many people it returns. Counts and filter lists use none.

```text
Use www.enrich.chat to list the VPs of Operations at northwindfreight.example.com, then find the work email for the first one and tell me its status and the credits used
```

## Which address should you add, /mcp or /mcp/oauth?

Add `https://enrich.chat/mcp` to try the tools with no account, or `https://enrich.chat/mcp/oauth` to use your own plan. Both serve the same 11 tools. The difference is who the app acts as.

**Figure 1: Two MCP addresses, one set of tools.** An AI app can connect to https://enrich.chat/mcp with no sign-in and a daily allowance of free lookups, or to https://enrich.chat/mcp/oauth by signing in with an Enrich.chat account and using that plan's credits. Both addresses reach the same 11 tools.

*The free address and the OAuth address reach the same tools. Source: [Enrich.chat install page](https://enrich.chat/install/).*

| Address | Sign-in | Whose usage | Use it when |
| --- | --- | --- | --- |
| `https://enrich.chat/mcp` | None | 10 free lookups a day | You want to try it, or an agent needs a quick lookup |
| `https://enrich.chat/mcp/oauth` | Your Enrich.chat account, over OAuth | Your plan's credits, or your own free lookups until you pick a plan | You use it every day, or want the app listed in your dashboard so you can disconnect it |

Until you pick a plan, a signed-in connection uses your account's free lookups. When you pick a plan on the [account page](https://enrich.chat/account/), the same connection uses the plan's credits. On a [team workspace](https://enrich.chat/features/team-workspaces/), those credits are the team's shared pool, and a viewer's connection gets only the free allowance. If an app cannot do OAuth, `/mcp` also accepts an API key as an `Authorization: Bearer` header. Keys are optional and come with a paid plan.

## How does the OAuth sign-in work?

When an app first calls `/mcp/oauth`, the server answers 401 and points to its sign-in rules. The app registers itself, sends you to Enrich.chat to log in and press Allow, and trades the result for a token. The steps follow the [MCP authorization specification](https://modelcontextprotocol.io/specification/2026-07-28/basic/authorization), which builds on OAuth 2.1.

**Figure 2: The OAuth sign-in flow for the Enrich.chat MCP server.** Six steps between an AI app and Enrich.chat. The app calls the OAuth address and gets a 401 with a link to the rules, reads the rules and registers, opens the sign-in page with a PKCE challenge, the person logs in with a 6-digit code and presses Allow, the app is redirected back with a one-time code, sends the code and PKCE verifier to get an access token, and then calls tools with the token.

*The app asks, you allow, and the app gets a token. The RFC numbers name the standard behind each step.*

1. **Discovery.** The app calls `https://enrich.chat/mcp/oauth` with no token. The server returns 401 with a `WWW-Authenticate` header that points to its [protected resource metadata](https://www.rfc-editor.org/rfc/rfc9728), which the specification requires of MCP servers. A second document, the [authorization server metadata](https://www.rfc-editor.org/rfc/rfc8414), lists the endpoints and says the server accepts PKCE with the S256 method.
2. **Registration.** The app registers itself through [dynamic client registration](https://www.rfc-editor.org/rfc/rfc7591) and receives a client id. Enrich.chat accepts public clients only, so there is no client secret to store.
3. **Sign-in.** The app opens Enrich.chat in your browser. You log in with your email and a 6-digit code, with no password. The page names the app and says it can find work emails, mobile numbers, people and companies and spend your credits. It also says the app cannot see your account details, API keys or billing. Press Allow to continue.
4. **Token.** The app trades the one-time code and its secret PKCE verifier for an access token. [PKCE](https://www.rfc-editor.org/rfc/rfc7636) means a stolen code is useless without the verifier, and the [OAuth 2.1 draft](https://datatracker.ietf.org/doc/html/draft-ietf-oauth-v2-1-13) makes it mandatory for every client. The [MCP security rules](https://modelcontextprotocol.io/specification/2026-07-28/basic/authorization/security-considerations) also tell clients to use S256 and to stop if the server does not advertise PKCE.

The token is sent as `Authorization: Bearer` on every call, and the tools run as you. That is why the OAuth address spends your plan's credits and the free address does not.

## How do you add it to Claude, ChatGPT, Cursor, VS Code, Claude Code and Codex?

Paste the server URL into your app's connector or MCP settings, then sign in when asked. Use `https://enrich.chat/mcp/oauth` for your plan and `https://enrich.chat/mcp` for no sign-in. Menu names change between releases, so each row links to the vendor's current steps.

| App | Where to add it | What to enter |
| --- | --- | --- |
| Claude | Customize, Connectors, then Add custom connector, per [Claude's help article](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp). On Team and Enterprise, an owner adds it first. | The URL. Choose Sign in now for the OAuth address, or No sign in for the free one. Claude's Free plan allows one custom connector. |
| ChatGPT | Plugins page, plus button, Add custom MCP server, per [OpenAI's guide](https://developers.openai.com/apps-sdk/deploy/connect-chatgpt). Workspace policy may limit this. | A name, the URL, and OAuth when it asks about authentication. |
| Cursor | `~/.cursor/mcp.json`, per [Cursor's docs](https://cursor.com/docs/mcp). | `{"mcpServers": {"enrich": {"url": "https://enrich.chat/mcp/oauth"}}}` |
| VS Code | MCP: Add Server, then HTTP, or `.vscode/mcp.json`, per [VS Code's docs](https://code.visualstudio.com/docs/copilot/customization/mcp-servers). | `{"servers": {"enrich": {"type": "http", "url": "https://enrich.chat/mcp/oauth"}}}`. VS Code [tries dynamic client registration first](https://code.visualstudio.com/api/extension-guides/ai/mcp). |
| Claude Code | Your terminal, per [Anthropic's docs](https://code.claude.com/docs/en/mcp). | `claude mcp add --transport http enrich https://enrich.chat/mcp/oauth`, then run `/mcp` to sign in. |
| Codex | Your terminal, per [OpenAI's Codex docs](https://learn.chatgpt.com/docs/extend/mcp?surface=cli). | `codex mcp add enrich --url https://enrich.chat/mcp/oauth`, then `codex mcp login enrich`. |

The [install page](https://enrich.chat/install/) also has one-click links for Cursor and VS Code on the free address. If you want a prompt your agent can follow, see [contact data for AI agents](https://enrich.chat/agents/). For context on where this fits, read [the best MCP servers for sales](https://enrich.chat/blog/best-mcp-servers-for-sales/) and [how sales teams use Claude](https://enrich.chat/blog/claude-for-sales/).

## How do you see and disconnect connected apps?

Open the dashboard at [Connect an agent](https://enrich.chat/app/#/agent). The Connected apps list shows every connected app with the date it connected and when it was last used. Press Disconnect and the app stops working at once. You can connect it again later.

**Figure 3: The Connect an agent screen in the dashboard.** The Connect an agent screen shows the OAuth server address with a copy button, a Connected apps table with the app name, when it connected, when it was last used and a Disconnect button on each row.

*The Connect an agent screen in the [dashboard](https://enrich.chat/app/). App names and dates are illustrative.*

The Usage tab in the same screen covers the dashboard, the API and connected AI apps together, so you can see what an app spent. Keys and billing sit on the [account page](https://enrich.chat/account/).

## What are the limits of the Enrich.chat MCP server?

The server finds and counts contacts. It does not act on them. Know these limits before you build a workflow on it.

- **Lookups only.** The 11 tools do not save to lists, run [bulk enrichment](https://enrich.chat/features/bulk-enrichment/), start [background jobs](https://enrich.chat/features/background-jobs/) or edit your [Do not contact list](https://enrich.chat/features/do-not-contact/). Those live in the dashboard and the API, where [match tiers](https://enrich.chat/features/match-tiers/) also run. See [lists and export](https://enrich.chat/features/lists-and-export/).
- **No sending.** There is no email sending, sequencing or dialer. The server finds the contact. Another tool sends the message.
- **A page of 25.** A search returns up to 25 people per call and up to 1,000 per search. Narrow the filters to go deeper. Keep to about one call a second.
- **Remote only.** It is an HTTPS server that answers each call with one JSON reply. There is no local package and no streaming.
- **Dynamic registration.** Enrich.chat registers apps with RFC 7591. The current MCP specification marks it [deprecated, kept for backwards compatibility](https://modelcontextprotocol.io/specification/2026-07-28/basic/authorization/client-registration) in favor of Client ID Metadata Documents, which Enrich.chat does not support today. Apps that register dynamically still connect.
- **Inputs.** Company lookups take a website, not a name. Work emails only on the email input. Treat every contact as a strong lead, not a guarantee, and read the status on each field.

To build on the same lookups with code, see the [API page](https://enrich.chat/features/api/) and the [API reference](https://enrich.chat/docs/). For agents that run outreach research end to end, see [AI SDR agents](https://enrich.chat/use-cases/ai-sdr-agents/). Plans are on the [pricing page](https://enrich.chat/pricing/).

## Frequently asked questions

### What is an MCP server for contact data?

An MCP server for contact data is a remote service that gives an AI app lookup tools through the Model Context Protocol. Enrich.chat's server has 11 tools for people, work emails, mobile numbers, companies, tech stacks and email formats, and its search tools take extra filters through `more_filters`. You add its URL once, and Claude, ChatGPT or Cursor can call the tools from a normal chat. The [MCP overview](https://modelcontextprotocol.io/docs/getting-started/intro) explains the standard itself.

### Does the Enrich.chat MCP server need an API key?

No. The address `https://enrich.chat/mcp` needs no key and no account, and gives 10 free lookups a day. The address `https://enrich.chat/mcp/oauth` signs you in with your Enrich.chat account over OAuth, so there is no key to copy there either. API keys are optional and come with a paid plan. `/mcp` also accepts one as an `Authorization: Bearer` header if an app cannot use OAuth.

### What is the difference between /mcp and /mcp/oauth?

Both addresses serve the same 11 tools. `/mcp` needs no sign-in and uses 10 free lookups a day. `/mcp/oauth` sends you to Enrich.chat to log in and press Allow, then uses your plan's credits, or your own free lookups until you pick a plan. On a team workspace, the plan is the team's shared pool. An OAuth connection also shows up in the Connected apps list in your dashboard, where you can disconnect it.

### Can I use Enrich.chat as a Claude or ChatGPT connector for lead data?

Yes. In Claude, add a custom connector with the server URL, as [Claude's help article](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp) describes. In ChatGPT, add a custom MCP server, as [OpenAI's guide](https://developers.openai.com/apps-sdk/deploy/connect-chatgpt) describes. Your plan or workspace may limit custom connectors, and menu names change, so follow each vendor's current steps. The [install page](https://enrich.chat/install/) lists the URLs to paste.

### How much does the Enrich.chat MCP server cost?

The server is free to add. The first 10 lookups each day cost nothing and need no signup, and a lookup that finds nothing uses 0 credits. After that, credits come from a monthly plan: Growth $29 for 100,000 credits, Scale $90 for 500,000 credits, Pro $299 for 2,500,000 credits. A work email found uses 10 credits, a mobile 250 and a profile 5. The [pricing page](https://enrich.chat/pricing/) has a calculator.

### How do I disconnect an app from the Enrich.chat MCP server?

Open [Connect an agent](https://enrich.chat/app/#/agent) in the dashboard, find the app under Connected apps, and press Disconnect. The app stops working at once, and you can connect it again later. An app that uses `/mcp` without signing in has no connection to remove, so delete the connector inside the app instead.

## Sources

1. [MCP authorization specification (2026-07-28)](https://modelcontextprotocol.io/specification/2026-07-28/basic/authorization)
2. [MCP client registration (2026-07-28)](https://modelcontextprotocol.io/specification/2026-07-28/basic/authorization/client-registration)
3. [MCP authorization security considerations (2026-07-28)](https://modelcontextprotocol.io/specification/2026-07-28/basic/authorization/security-considerations)
4. [MCP tools specification (2026-07-28)](https://modelcontextprotocol.io/specification/2026-07-28/server/tools)
5. [What is the Model Context Protocol?](https://modelcontextprotocol.io/docs/getting-started/intro)
6. [RFC 9728: OAuth 2.0 Protected Resource Metadata](https://www.rfc-editor.org/rfc/rfc9728)
7. [RFC 8414: OAuth 2.0 Authorization Server Metadata](https://www.rfc-editor.org/rfc/rfc8414)
8. [RFC 7591: OAuth 2.0 Dynamic Client Registration Protocol](https://www.rfc-editor.org/rfc/rfc7591)
9. [RFC 7636: Proof Key for Code Exchange (PKCE)](https://www.rfc-editor.org/rfc/rfc7636)
10. [The OAuth 2.1 Authorization Framework (draft 13)](https://datatracker.ietf.org/doc/html/draft-ietf-oauth-v2-1-13)
11. [Get started with custom connectors using remote MCP](https://support.claude.com/en/articles/11175166-get-started-with-custom-connectors-using-remote-mcp)
12. [Connect Claude Code to tools via MCP](https://code.claude.com/docs/en/mcp)
13. [Connect your MCP server to ChatGPT](https://developers.openai.com/apps-sdk/deploy/connect-chatgpt)
14. [Model Context Protocol in Codex](https://learn.chatgpt.com/docs/extend/mcp?surface=cli)
15. [Model Context Protocol (MCP) in Cursor](https://cursor.com/docs/mcp)
16. [Add and manage MCP servers in VS Code](https://code.visualstudio.com/docs/copilot/customization/mcp-servers)
17. [MCP developer guide for VS Code](https://code.visualstudio.com/api/extension-guides/ai/mcp)

## Examples

**Claude Code**

```text
claude mcp add --transport http enrich https://enrich.chat/mcp/oauth

# Then start Claude Code, run /mcp, choose enrich,
# log in to Enrich.chat and press Allow.

# No account? Use the free address instead:
#   claude mcp add --transport http enrich https://enrich.chat/mcp
```

**Cursor**

```json
{
  "mcpServers": {
    "enrich": {
      "url": "https://enrich.chat/mcp/oauth"
    }
  }
}

// Save as ~/.cursor/mcp.json. Use /mcp for no sign-in.
```

**Tool call**

```text
Ask: Who is the CFO of northwindfreight.example.com?

tool: company_people
arguments: {
  "domain": "northwindfreight.example.com",
  "title": "CFO",
  "limit": 1
}
```

**Result**

```json
{
  "found": true,
  "count": 1,
  "domain": "northwindfreight.example.com",
  "people": [{
    "id": "per_7Qm3xB2nKd9TfVhC",
    "name": "Maya Okafor",
    "title": "CFO",
    "company": "Northwind Freight",
    "has_email": true,
    "has_mobile": false
  }],
  "credits_used": 5,
  "free_lookups_remaining": 9
}

// Illustrative. Fictional person and company.
```

## Prices

| What you get | Credits |
| --- | --- |
| Work email found | 10 |
| Mobile number found | 250 |
| Company tech stack or profile found | 5 |

10 free lookups a day for every IP address, no signup. Nothing is charged when nothing is found. Monthly plans start at $29 for 100,000 credits.

## Try this prompt

Look up a contact from your AI app

```text
Use www.enrich.chat to list the VPs of Operations at northwindfreight.example.com, then find the work email for the first one and tell me its status and the credits used
```

## Related

- [Contact and company data API](https://enrich.chat/features/api/)
- [Data for AI SDR agents](https://enrich.chat/use-cases/ai-sdr-agents/)
- [People search](https://enrich.chat/features/people-search/)