“Which open cases have a statute of limitations date in the next 90 days?”
search_case_fieldsScans the SOL field across every open matter and returns the list, nearest deadline first.
MCP for case management
Your case files, in the AI you already use.
Connect your case management system once. CaseMCP gives you a URL. Paste it into Claude or ChatGPT and ask about your cases in plain English. No IT project, no sync, no new tab to live in.
Read-only by default. Rotate or revoke anytime. A paid product, set up with us on a call.
https://casemcp.com/c/k4x…9rq2/mcpWorks inClaudeChatGPTAny MCP client
Setup is a paste.
1
Connect
On a short call we connect your case management system with your own credentials. Your permissions, encrypted at rest.
2
Copy
CaseMCP reads your field schema and hands you one private URL, scoped to your firm and revocable anytime.
3
Paste
Drop it into Claude or ChatGPT as a connector. Ask your first question about a real case within the minute.
What firms ask.
Not a chatbot bolted onto a portal. Your actual AI assistant, with your actual cases in reach. Whatever seat you sit in.
“Brief me on the Ramirez case before my 2pm call.”
get_case_briefOne tight summary: posture, treatment status, demand history, and what moved since you last looked.
“Who is the adjuster on Patel, and where did our last demand land?”
get_casesearch_case_fieldsAdjuster name and contact straight from the file, plus the demand history, so you walk into the negotiation current.
“Pull the depo transcripts and medical records from Alvarez.”
get_case_documentsget_document_download_urlsFinds the records in the file and hands you download links. No digging through folders.
The difference
It knows what your fields are actually called.
Every case management system is configured per firm. Your project types, sections, and field keys exist nowhere else, so a generic integration has to guess at them and guesses wrong.
CaseMCP reads your firm's real schema when you connect and hands it to the AI. It asks for the right field, by its right name, the first time.
describe_schemayour firm
Personal Injury – Auto
- solDateStatute of Limitationsdate
- medsSummaryMedical Summarytext
- demandAmountDemand Amountcurrency
- adjusterInfoAdjustercontact
- treatmentStatusTreatment Statusdropdown
A small, sharp surface.
Thirteen read tools cover the questions firms actually ask. Writes exist, but they ship off and stay off until you flip them on per workspace.
describe_schemasearch_casesget_caseget_case_briefget_case_activityget_case_documentsget_document_download_urlsget_case_teamget_case_sectionssearch_case_fieldssearch_contactsget_contactget_contact_metadata+ 8 write tools, opt-inFilevine, Neos, and Clio today.
The connector model is the same for every case management system: your credentials, your schema, one URL. Three are live today and the next systems are in the works.
Filevinelive
Neoslive
Cliolive
MyCasecoming
Smokeballcoming
CasePeercoming
Litifycoming
Your cases stay yours.
Your credentials, encrypted
Your CMS credentials are encrypted at rest and used only to answer your own requests. They carry the permissions of the person who created them, nothing more.
Case data is never stored
We keep the connection and a cache of your field names. Every answer is fetched live from your case management system and passed straight through.
Read-only by default
Out of the box the AI can look and never touch. Write access is a separate, explicit switch per workspace.
Revoke in one click
The URL is a bearer secret, so we treat it like one. Rotate it or delete the connection from the dashboard and it stops answering immediately.
Five minutes to your first answer.
We set you up on a short call, then you paste one URL into your AI and ask it something you would have paid a case manager an hour to look up.