■ Model Context Protocol
Call Zurvek from inside your assistant
MCP is the open standard that lets an assistant use outside tools. Once ours is live, you can screen a company or pull a verdict trail without leaving Claude, ChatGPT or Cursor.
Connect once
Add Zurvek as a connector in Claude, ChatGPT, Cursor or any MCP client. One authorisation, no keys to copy around.
Ask in plain language
Ask your assistant to screen a company or pull a verdict. It calls Zurvek and returns the result inside the conversation.
Same engine, same verdict
There is no second model and no lighter version. MCP calls run the same adversarial pipeline as the dashboard.
■ Planned tools
screen_companyRun an adversarial screen from a company name or URL.
Plannedget_verdictPull a completed verdict trail by evaluation id.
Plannedlist_evaluationsList your recent evaluations and their status.
Plannedget_memoRetrieve the IC memo for a completed evaluation.
PlannedAt launch: read and screen operations only. No billing actions, no admin actions, no bulk data export over MCP.
Access and safety
- Connections are per user. The assistant signs in as you and sees only what your account can already see.
- Nothing becomes public. No other account's evaluations are reachable through your connection.
- Tier limits and quota rules apply identically over MCP.
Tell me when MCP opens
Pricing for MCP access follows the existing tiers. See clearance tiers →
