Lumonic

Query private-credit portfolio data — borrower financials, covenant status, reporting deliverables — from an AI agent.

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 Lumonic connector does

The Lumonic MCP server connects a private-credit or private-equity portfolio monitoring workspace to Claude, ChatGPT and any MCP-compatible agent. Anthropic's directory lists 24 tools that read borrower financials, covenant status, reporting deliverables and schema metadata. Every one reads; none writes. Sign-in is OAuth, and the server requests a single scope named read.

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.

Lumonic tools (24)

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

Limits

  • Lumonic publishes no tool list. The 24 names come from Anthropic's directory. We searched Lumonic's help centre, its /mcp, /mcp/library and /mcp/pitchbook pages and its connector guide, and not one of the 24 tool names appears as a literal string on any Lumonic page. This is the "no vendor enumeration" case: there is no second list to diff against, so we cannot report a clean listing — only that nothing exists to check it with. That is a weaker statement than "no undisclosed tools were found".
  • What partly substitutes for an enumeration is a negative one. Lumonic's /mcp page and its connector guide both constrain the surface to read access, and Anthropic's directory description uses only read verbs — discover, build, summarize, identify, validate, generate, inspect. A constraint is more useful than a feature list, but it names no tools, so renames would be invisible to us either way.
  • We could not read tool schemas or safety annotations. The endpoint returned 401 to an anonymous request, so no parameter-level detail is published here.
  • One prompt is listed and we could not read it. Anthropic's directory lists a build_lumonic_workbook prompt. prompts/list requires authentication, so we cannot say what it contains or what it calls.
  • Lumonic's own contract terms were unreadable. /footer/license-agreement and /footer/terms-of-use returned byte-identical responses on 2026-08-23, both Framer shells whose HTML contains no legal prose and no embedded JSON payload. A control path under the same prefix returned a genuine 404, so this is not a blanket catch-all — those two routes specifically do not serve their documents to an unauthenticated fetch. We therefore report no contractual terms for this connector, and the confidentiality analysis above rests on the nature of credit agreements generally, not on Lumonic's paper.
  • reporting_reminder_targets is unresolved. We could not establish from documentation whether it only identifies who is delinquent or also dispatches a reminder to them, and we did not test it. Lumonic does send notification emails.
  • "Currently read-only" is a statement about today. Lumonic's own wording is that the connector "is currently read-only". The single read scope is consistent with that, but a scope named read is a label, not an enforcement proof — we could not inspect the server's dispatch path.
  • Permission inheritance is a vendor claim we could not verify. Lumonic states MCP queries run as the requesting user's role. We had no credentials and did not test it.
  • The connector is for Lumonic customers only. Lumonic states the MCP is included for every Lumonic customer; there is no standalone or trial access.
  • Accuracy is your problem to check. Lumonic's troubleshooting section anticipates the model using the wrong metric and returning incomplete results, and advises asking it to confirm the workspace context, reporting period, entities and fields it used. Ask for source dates and traceback links when the answer will be acted on.

Frequently asked questions

Can the Lumonic connector change a reported figure or mark a covenant breach?

No. All 24 tools read. Lumonic states its connector is currently read-only, so an analyst querying a borrowing base cannot modify the underlying record. The platform does let staff edit restatements and set compliance status in its own interface, but no MCP tool reaches either. The OAuth grant carries one scope, read, which matches.

What borrower data can the Lumonic MCP server read?

Third-party company financials and loan economics. Lumonic states the connector gives models read access to financials, cash flows, covenants, reporting status and entities. Its query language documents fields including adjusted_ebitda, total_debt_outstanding, accounts_receivable and monthly-burn. That is your borrowers' confidential reporting, not your own firm's data.

Does connecting Lumonic to Claude breach a credit agreement NDA?

Lumonic does not address that question, and you should not assume it does. Its documentation covers model training and permissions, not the confidentiality clauses in your credit agreements. Those clauses govern disclosure of borrower information to third parties, and an AI provider is a third party. Ask counsel before connecting a portfolio.

Does Lumonic use my portfolio data to train an AI model?

Lumonic states it does not. It says it does not contribute client data to any model's training and passes queries through under enterprise terms that exclude training. Lumonic also notes the model provider may log conversations under its own terms, which it calls a separate decision from what the data platform retains.

Does each Lumonic MCP query cost money or consume credits?

No. Lumonic states the MCP is included for every Lumonic customer, and no documentation mentions per-query pricing, credits or metered usage. No tool reports a balance or quota either, so nothing in the surface could observe spend — but with no per-call charge documented, there is nothing to observe.

Does the connector respect each user's Lumonic permissions?

Yes, according to Lumonic. It states each MCP query maps to the requesting user's existing role, so the model inherits the permission boundaries your team already administers. Lumonic's permission model runs Cluster to Namespace to Entity with allow and deny rules, and deny always wins. We could not verify this behaviour ourselves.

How many tools does the Lumonic MCP server have?

Anthropic's directory lists 24, and Lumonic publishes no tool list of its own. The endpoint returned 401 to an anonymous request, so we never read a live tools/list and cannot confirm the count independently. Treat 24 as Anthropic's figure rather than a verified total.

Why does Lumonic not appear in my Claude connector list?

Your workspace owner has probably not added it yet. Lumonic states that on Claude Team or Enterprise an Owner or Primary Owner may need to enable the connector before individual users can connect. Lumonic documents a fallback: copy your MCP URL from Settings, open the MCP tab, and add it as a custom connector.

Sources

More Finance connectors

  • Affinity

    Search your CRM, prep for meetings and update deal records from your AI assistant.

  • Aiera

    Pull earnings-call transcripts, event calendars, SEC filings and company publications into your AI assistant.

  • Airwallex Developer

    Search Airwallex documentation and exercise a sandbox account 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 Lumonic to work.

Connect Lumonic 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.