Integrations · Claude (Anthropic)
RichAPI runs a hosted MCP server. Connect it to Claude once, then ask for enrichment in plain English. Every tool call returns an execution_log receipt showing which providers ran and what you were charged.
Last updated September 22, 2026
Create a RichAPI account at app.richapi.ai/signup and copy your API key.
25 free credits, no card.
Add the RichAPI server to your Claude MCP config.
Claude Desktop and Claude Code go through the mcp-remote shim; the server URL is https://mcp.richapi.ai/mcp and auth is an x-api-key header.
Restart Claude.
The RichAPI tools appear in the tool list — same endpoints as the REST API, same key, same credit pool.
Ask for something.
'Find the work email for the VP of Engineering at stripe.com.' Claude picks the tool, runs it, and shows you the receipt.
Copy, paste, edit the brackets, run.
Why it matters
Enrichment normally means a Clay table, a CSV, or a script. Through MCP it is a sentence in a chat window, and the tool call comes back with a machine-readable record of which providers ran and what each one returned.
RichAPI runs a hosted MCP server. You connect it to Claude once, and from then on enrichment is something you type in a sentence instead of something you build. There is no package to install and nothing to run locally. The server lives at `https://mcp.richapi.ai/mcp`, authenticates with your API key in an `x-api-key` header, and exposes the same endpoints as the REST API at `api.richapi.ai`, on the same key and the same credit pool. Whether Claude calls it or your code does makes no difference to your account.
Create an account at `app.richapi.ai/signup` and copy your key. Then add the server to your Claude MCP config. Claude Desktop and Claude Code reach a remote server through the `mcp-remote` shim: ```json { "mcpServers": { "richapi": { "command": "npx", "args": [ "mcp-remote", "https://mcp.richapi.ai/mcp", "--header", "x-api-key: ${RICHAPI_KEY}" ] } } } ``` Restart Claude and the tools show up. Client config formats change more often than this page does — the live OpenAPI at https://api.richapi.ai/api/v1/openapi.json is canonical if the two disagree. An in-app connector path (Settings → Connectors) is not live yet. Use the config above.
The tool names map one-to-one onto the REST endpoints, so anything in the API is available in the chat. The useful ones for a GTM engineer: - `enrich_company` — company data from LinkedIn: size, industry, specialties, employee insight. - `enrich_profile` — a LinkedIn profile: headline, location, experience, education. - `email_finder` — a work email from a name plus domain, or a LinkedIn URL. - `email_verifier` — deliverability on an address you already have. - `phone_finder` — a mobile from a LinkedIn URL, or name plus company. - `people_search` — profiles filtered by domain, industry, headcount, title, location. - `post_keyword_search` — who posted about a topic recently, which is how you find people who are in-market this week rather than in a database from last year. - `web_tech_stack` and `website_intelligence` — what a domain runs. Some prompts that do real work: > Find a verified work email for the VP of Engineering at acme.com. If it misses, tell me which providers were tried and what each returned. > Take these 30 LinkedIn company page URLs. Enrich each company, pull the head of marketing, and give me a table with headcount, industry and whether we found a verified email. Total the credits at the end. > Verify this list before I send. Split it into valid, risky and invalid, and tell me what the verification cost. Claude chooses the tool, fills the arguments from your prose, and runs it. You did not write a request body, and you did not open a spreadsheet.
Every waterfall call comes back with an `execution_log`: which providers ran, in what order, and what each returned. On a miss the response also carries `"billed": false`. The exact shape is a captured sample, not something we reconstruct on a marketing page. We do not publish the provider roster, because it changes as we add and drop providers. The `execution_log` is the honest version of that answer: instead of a list on a marketing page, you get the actual per-call record of who ran on your request. Inside Claude this changes what a failure looks like. When an assistant returns nothing, you cannot tell a real miss from a broken call. When it returns a log of four attempts, three with no data and one that timed out, you know whether to retry, change the input, or stop.
Billing runs on prepaid credits from one pool, shared between MCP and REST. Per-call rates for the tools above: | Tool | Cost | | --- | --- | | `enrich_company` | `1 credit` | | `enrich_profile` | `1 credit` | | `email_finder` | `5 credits` | | `email_verifier` | `2 credits` | | `phone_finder` | `25 credits` | | `people_search` | `0.1 credits per result` | On the waterfall endpoints — `email_finder`, `email_verifier`, `phone_finder`, — a call that finds nothing costs zero credits. Every provider was paid, you were not. That is scoped deliberately: the other endpoints bill on a successful 2xx response like any normal API, and we would rather say so here than have you find out on an invoice.
Every RichAPI tool is a read-only lookup. Nothing in the catalog sends an email, posts anything, or writes to your CRM. The worst outcome of a badly phrased prompt is spent credits, not a message to a customer with an unfilled merge field in it. The tool catalog is served from the live endpoint config rather than baked into a release, so new endpoints appear in your client without you doing anything.
We are the data layer, not a workflow suite. There is no prospecting UI, no sequencer, no campaign tool, no CRM app. If you want tables you can click, use Clay. What you get here is one key, one credit pool, a set of tools an assistant can call, and a receipt for each call.
**Do I need a credit card?** No. 25 free credits on signup, no card. **Does it work in Claude Code as well as Claude Desktop?** Yes. Same config, same shim. **Can Claude spend my credits without me noticing?** It can spend them, which is why every result carries its own cost line and why the free tier exists. Try it on the 25 credits before you fund anything. **Is this an official Anthropic integration?** No. It is a standard MCP server, which Claude supports. **Can I use the same key from code?** Yes. Same key and same credits against `api.richapi.ai`. Prove a workflow in chat, then automate the version that worked.
25 free credits, no card. Connect the server and ask Claude for a verified email in the next two minutes. The endpoints behind the tools you just connected: [email finder](/api/email-finder), [email verifier](/api/email-verifier), [phone finder](/api/phone-finder), [person enrichment](/api/person-enrichment) and [company enrichment](/api/company-enrichment). For a worked example, see [agent-driven account research](/use-cases/agent-driven-account-research); for the same chain over hundreds of rows, [list build and verify](/use-cases/list-build-and-verify). Other MCP clients are under [integrations](/integrations). Rates are on [pricing](/pricing).