ZoomInfo

Search companies, find contacts and enrich B2B records.

Add to Agentman Studio

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

Resources

What the ZoomInfo connector does

The ZoomInfo MCP server connects ZoomInfo's B2B contact and company database to Claude, ChatGPT and any MCP-compatible agent. It exposes 21 tools for searching companies, finding named individuals at them, enriching records with contact details, and reading buying signals. Sign-in is OAuth. Five tools charge credits per record, so cost is the safety axis here, not deletion.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

ZoomInfo tools (21)

  • search_contacts
  • enrich_contact
  • search_companies
  • enrich_company
  • enrich_company_signals
  • account_research
  • contact_research
  • conversation_intelligence
  • get_recommended_contacts
  • find_similar_contacts
  • find_similar_companies
  • browse_audiences
  • get_audience
  • browse_engagements
  • get_gtm_context
  • update_gtm_context
  • search_scoops
  • search_news
  • search_intent_signals
  • lookup
  • submit_feedback

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Five tools spend money and none of them is named like it. The three enrich tools and the two research tools are metered; the other 16 are free. A verb census reads all five as ordinary lookups. Enrich charges one credit per record returned, capped at 25 records per call.
  • Credit exhaustion arrives inside a 200 response. LIMIT_EXCEEDED is a per-record match status alongside ordinary misses, so a drained balance is indistinguishable from an empty result unless the agent reads the match status. The request-level PFAPI0011 is an HTTP 400.
  • The consent screen offers nothing meaningful to narrow. The resource advertises three application scopes and no read-only grant. ZoomInfo's platform documents seventeen scopes with read and manage splits — including api:entitlements:read and separate api:data:contact and api:data:company — but does not scope this MCP resource with them. Even those seventeen fuse the two operations that matter here: the OpenAPI spec describes each data scope as covering search and enrich, so the paid reveal cannot be granted separately from the free search.
  • Entitlement narrows responses silently. Field-level gating means a tool stays visible while returning less. The Lookup Enrich endpoint reports per-field access, and it is the only way to distinguish an unlicensed field from a genuinely empty one.
  • Rate limits run in three windows at once. Per-second, per-hour and per-day, each enforced independently; any one exhausted returns HTTP 429. ZoomInfo publishes 5 req/s on Builder, 25 on Standard and 35 on Scaling, with matching hourly and daily caps. Rejected requests do not consume quota. Search pagination is capped at 100 pages and 100 records per page.
  • The audiences documentation over-promises the MCP surface. ZoomInfo's audiences page says audience creation and management are supported "through both MCP and APIs" and names Create, Update and Delete operations for audiences, columns, rows and folders. Neither the MCP tools table nor Anthropic's tool list contains any of them — MCP gets browse_audiences and get_audience, both reads. Those write operations appear to be REST-only.
  • Model training is prohibited on this data. ZoomInfo requires model training to be disabled in your AI client before connecting.
  • The directory's documentation URL is dead. Anthropic's snapshot points at an MCP overview page that returns HTTP 404 and is absent from ZoomInfo's sitemap and documentation index.
  • We could not read tool schemas. The endpoint returned HTTP 401 to our anonymous handshake, so no inputSchema, no safety annotations, and no live prompt or resource listing appear here. Tool names come from Anthropic's directory snapshot; credit tiers, prerequisites, entitlement behaviour and limits come from ZoomInfo's own documentation.

Frequently asked questions

Does the ZoomInfo MCP server spend credits when an agent calls a tool?

Yes, and five of the 21 tools do it. ZoomInfo's tool table marks the three enrich tools as Bulk Data and the two research tools as AI Action, while search, lookup and browse are Free. Its API reference is more precise still: each record an enrich call returns charges one credit unless that record is already under management, and a no-match or an error charges nothing at all.

What happens when the ZoomInfo credit limit runs out mid-run?

The call still returns HTTP 200 and the exhausted records come back looking like misses. ZoomInfo's API reference documents LIMIT_EXCEEDED as a per-record match status inside the meta object, not as a request-level failure. An agent that only checks the status code will read a drained balance as an empty result, so read the per-record match status before concluding a person is not in the database.

Which ZoomInfo MCP tools can change something rather than read it?

Two of the 21. Update GTM Context writes your organisation's structured GTM configuration, and submit_feedback sends feedback to ZoomInfo. Everything else reads. The directory nevertheless labels the connector Read and write, which is accurate but coarse — the more useful split is metered versus free, because five tools that read nothing but your credit balance still cost money.

What OAuth scopes does the ZoomInfo MCP server ask for?

Seven, of which only three are application scopes: zi_api, zi_mcp and api:data:mcp. The other four are the OpenID identity set. Both the resource descriptor and the authorization server metadata publish the same seven, verified live. ZoomInfo's platform separately documents seventeen far narrower scopes with read and manage splits, but it does not offer them for this MCP resource.

Can I tell which ZoomInfo fields my account is actually licensed for?

Yes, and that is unusual for a data vendor. ZoomInfo's Lookup Enrich endpoint accepts a field-type filter of output and returns every available field along with whether your account has access to it. Entitlement here narrows the response rather than hiding tools, so without that check an unlicensed field looks identical to a person who simply has no phone number on record.

What personal data does the ZoomInfo connector return about people?

Business contact data about individuals who never gave it to you. ZoomInfo's own documentation lists name, work and supplemental email, phone, job title, seniority, employer, street address, postcode, education and years of experience as contact attributes, across a database it sizes at over 500 million professionals. Its privacy policy confirms it builds these profiles from sources other than the individuals themselves.

Can I use ZoomInfo MCP data to train an AI model?

No. ZoomInfo's connection guide states plainly that data accessed through MCP must not be used for AI model training, and it asks you to switch model training off in your AI client before connecting at all. That is a prerequisite rather than advice, and it sits alongside the licence requirement in the same before-you-begin list, so treat it as a condition of use.

Why does the ZoomInfo documentation link in Anthropic's directory not work?

The page it points at no longer exists. Anthropic's snapshot names an MCP overview URL that returns HTTP 404, and that slug appears in neither ZoomInfo's sitemap nor its machine-readable documentation index. The live equivalents are the Connect to ZoomInfo MCP and Available MCP Tools pages, both listed in the sitemap and both serving clean Markdown.

Sources

More Sales & CRM connectors

  • Abency

    Run a marketing agency's social, SEO, ads, WordPress and email from an AI agent.

  • Actively

    Read per-account sales intelligence — agent memory, decisions, signals — from an AI agent.

  • Adobe Journey Optimizer

    Inspect campaigns, journeys and channel configurations in Adobe Journey Optimizer from an AI agent.

Browse all MCP connectorsClaude and ChatGPT setup guideBrowse AI skills

Product names and logos belong to their owners. Agentman is not affiliated with a connector's maker unless the page says the connector is built and run by Agentman.

Use it in an agent

Put ZoomInfo to work.

Connect ZoomInfo once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Open in Agentman Studio
Call (650) 285-1019Our AI receptionist answers.