Skip to main content

​
What is MCP?

The Model Context Protocol (MCP) is an open standard that allows AI assistants to connect to external tools and data sources. Jarvi exposes an MCP server so you can use AI assistants to search, read, and update your Jarvi data directly.

​
Compatible clients

Any MCP-compatible client can connect to Jarvi, including:

​
Configuration

​
Claude

  1. In Claude, on the web or in the desktop app, open Customize > Connectors.
  2. Click + Add then Add custom connector.
  3. Enter the name Jarvi and the URL https://mcp.jarvi.tech, then click Continue and Add, keeping the default authentication settings.
  4. Log in with your Jarvi account in the window that opens (OAuth2).
  5. In a conversation, click + then Connectors and turn on Jarvi.
Team and Enterprise plans: an organization owner first adds the connector in Organization settings > Connectors (Add > Custom > Web). Each member then clicks Connect next to Jarvi in Customize > Connectors. Claude’s free plan allows a single custom connector.

​
ChatGPT

  1. In ChatGPT on the web, open Settings > Security and login and turn on Developer mode.
  2. In Plugins, click + and create an app pointing to https://mcp.jarvi.tech, with OAuth authentication.
  3. Authorize access with your Jarvi account.
  4. In a conversation, turn on Jarvi from the Developer mode menu.
Available on Plus, Pro, Business, Enterprise and Edu plans, on the web only. ChatGPT only uses OAuth: the private API key does not work with it.

​
Other clients

For clients that don’t support OAuth2, use a private API key:
  • Transport: Streamable HTTP
  • URL: https://mcp.jarvi.tech
  • Authentication: x-api-key header with your private API key
Your private key grants full access to your Jarvi data. Never share it publicly.

​
Available tools

Once connected, the following tools are available for your AI assistant:
All operations run with your permissions. Row-level security applies — you can only access data your account has access to.

​
Prompt Examples

Once connected, try asking your AI assistant:
  • “Who am I?”
  • “Search for React developers in Paris”
  • “Show me the details of project X”
  • “Create a profile for John Doe, developer”
  • “Position this candidate on project X”
  • “Set the status of Acme company to Qualified”
  • “What custom fields exist on profiles?”
  • “Log a call with John Doe: discussed his availability”
  • “Remind me to follow up with Acme next Monday”
  • “Update John Doe’s title to Senior Developer”

​
Limitations

  • The same usage limits apply, counted on cumulative execution time. If you need more, contact our support team.
  • Deletion operations are blocked — use soft delete (updating deletedAt) instead.
  • Responses that are too large are automatically truncated. Use limit arguments in your queries.
  • Destructive tools (mutations, campaign launches, message follow-ups) require confirmation from the AI assistant before execution.