Meridian AI

Search a Meridian AI deal workspace: deals, companies, people, meetings, emails, notes and documents.

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 Meridian AI connector does

The Meridian AI MCP server connects a Meridian workspace — the CRM and deal-management platform built for private equity, venture capital, investment banking and private credit — to Claude, ChatGPT and any MCP-compatible agent. All 13 tools read: deals, companies, people, meetings, emails, notes, documents and reports. Sign-in is OAuth through your firm's existing SSO.

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.

Meridian AI tools (13)

  • get_report_data
  • list_filter_attributes
  • search_companies
  • search_deals
  • search_document_content
  • search_documents
  • search_emails
  • search_interactions
  • search_meetings
  • search_notes
  • search_people
  • search_reports
  • whoami

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

Limits

  • No application scopes exist. The server's RFC 9728 descriptor and its authorization server both advertise exactly openid, profile, email and offline — identity claims plus refresh tokens. Nothing at the OAuth grant separates reading deals from reading email or document text. Since every tool reads, no boundary is missing between a routine write and an irreversible one; the missing boundary is between one *kind* of confidential read and another.
  • We could not read tool schemas or safety annotations. The endpoint returned 401 to an anonymous request, so no parameter-level detail, readOnlyHint or destructiveHint is published here. The tool count and names come from Anthropic's directory, corroborated by Meridian's own documentation.
  • The read-only claim is the vendor's, not our observation. Meridian states the tools cannot modify your data, and the 13 names agree. We did not and would not call a tool to confirm it.
  • Permission inheritance is undemonstrated from outside. Meridian documents identity-bound requests and per-firm token isolation. Both are claims about behaviour inside an authenticated server that no external probe can check.
  • Rate limits are tight for bulk work. 4 requests per second per user with bursts to 10. Some of Meridian's own sample prompts — screening all companies on revenue, EBITDA margin and headcount growth, or mapping a whole portfolio — imply many chained calls, and the ceiling applies to all tool calls together.
  • Two Meridians share the directory. The Meridian Connector for QuickBooks is a different vendor's product with 64 tools including deletes. Confirm the endpoint host before acting on any tool list you find elsewhere.
  • The connector is new to partner tier. It was listed at community tier in our 2026-08-16 directory snapshot and appears at partner on 2026-08-23, with the same 13 tools, endpoint and documentation URL. The tier changed; the surface did not.
  • Meridian's llms.txt is served as RTF, not plain text. The file at www.meridian-ai.com/llms.txt returns HTTP 200 with Content-Type: text/plain but its body is RTF markup with escape codes. A machine reading it as the specification intends gets control sequences mixed into the prose. It is a publishing bug, not a content gap — the content is good.
  • No llms-full.txt and no Markdown route. Both return 404. The nominated documentation URL is the richest route here, which is the minority case.
  • No MCP app widget. The directory entry declares no has_mcp_app, so there is no client-side widget surface beyond the tools — the fifth surface a tool audit cannot reach does not apply.
  • No prompts or resources are advertised. The directory lists no prompt names. We could not probe prompts/list or resources/list because the handshake is gated, so treat this as unknown rather than as a confirmed absence.

Frequently asked questions

Can the Meridian AI connector change anything in my CRM?

No. Meridian's documentation states the tools are designed to read and compose and that they cannot modify your Meridian data. All 13 names carry a read verb — nine search, two get or list, one whoami, one report read. No name creates, updates, deletes, sends or shares anything, and no tool takes an operation or method as an argument.

Is this the same Meridian as the QuickBooks connector in Claude's directory?

No, and Anthropic lists both. Meridian AI is a private-markets CRM by Meridian AI Inc. of Miami. The Meridian Connector for QuickBooks is a separate accounting connector built by Pilot.com on a different host, with 64 tools that do write. Google's Meridian marketing-mix-modelling library is a third unrelated thing entirely.

What OAuth permissions does the Meridian AI MCP server request?

Four, and none of them narrow anything. The server's RFC 9728 descriptor advertises openid, profile, email and offline — three OpenID identity claims plus refresh-token issuance. No scope names a Meridian object or a verb, so consent grants the whole read surface at once. There is no partial grant covering deals but not emails.

What data can the Meridian connector read from my firm?

Your firm's confidential deal material. The tools reach deals, companies, people, meetings, emails, notes, documents and saved reports, including semantic search inside document text — CIMs and IC memos. Meridian states the agent sees exactly what you see, scoped to your organization, respecting private meetings, note access and email visibility rules.

Does Meridian AI MCP respect my existing permissions in the app?

Meridian states it does. Its documentation describes every request as identity-bound, running as you with the same organization scope and permissions you hold in the app, and says a token issued for one firm can never read another firm's data. We could not test that boundary — the endpoint returns 401 to an anonymous request — so it is Meridian's claim, not our observation.

How do I add Meridian AI to Claude?

In two stages. An administrator opens Organization settings, then Connectors, clicks Add, chooses Custom then Web, and enters the server URL https://mcp.meridian-ai.com/mcp. Each user then opens Customize, then Connectors, clicks Connect on the Meridian entry, enters their work email and approves access. Meridian suggests asking who am I in Meridian to confirm.

Are there rate limits on the Meridian AI MCP server?

Yes, per authenticated user rather than shared across your firm. Meridian documents a token-bucket model allowing 4 requests per second across all tool calls, with short bursts of up to 10. Exceeding it returns a standard MCP rate-limit error, and Meridian's advice is to wait a moment and retry rather than back off on a schedule.

Does calling a Meridian AI tool cost money or spend credits?

Meridian documents no per-call charge, no credit balance and no metered usage for the MCP server — only a rate limit. No tool among the 13 names a credit, balance, quota or billing operation either, so an agent could not check a cost even if one existed. Access is a function of your firm's Meridian subscription.

Sources

More Analytics connectors

  • Aleph

    Query live FP&A data, build reports and update dashboards from your AI assistant.

  • august

    Interview simulated customer twins, run focus groups and surveys from an AI agent.

  • ChartMogul

    Query MRR, ARR, churn, retention, cohorts, customers and sales pipeline from your AI assistant.

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 Meridian AI to work.

Connect Meridian AI 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.