Grok integration

Connected through MCP, Grok can call Omentir tools for workspace context, My Product, classic lead finders, Steal Customers agents, lead inspection, activity, scheduled sends, and approved work on existing conversations. LinkedIn credentials stay inside Omentir.

This is the same hosted MCP surface Claude and ChatGPT use for chat-style connectors, with workspace approval instead of pasting a long-lived token into the chat transcript.

Finish Omentir basics: account, LinkedIn connection, My Product. In Grok, add a custom connector with the Omentir MCP URL. Sign in on Omentir and approve Connect workspace. Enable tools in the conversation if prompted.

• MCP URL: https://omentir.com/api/agent/v1/mcp
• No API key required for the chat connector path
• List agents before create_agent to avoid duplicates
• Optional operator prompt from /for-agents as the first message

Ask Grok to run get_context and get_stats, confirm readiness, then propose a single agent configuration for your approval. Prefer one clear segment over a broad autonomous expansion of titles and geographies.

Daily send limits and campaign rules still apply. Grok is an operator interface, not a bypass around safety settings. You remain responsible for what gets sent from your LinkedIn account.

Give Grok the same written ICP you would give a teammate. If you only say 'find me SaaS founders,' you will get a wide, noisy list. If you name the role, the company size, and the trigger, the agent it creates will be easier to review.

Have Grok show the agent configuration before you leave it running. Confirm send windows and daily limits in the dashboard after. Chat operators are fast. The LinkedIn account still pays for a sloppy config.

If your team already standardized on Claude or ChatGPT, do not add Grok just to have another connector. One operator you actually watch is better than three you forget. Use Grok when that is already the chat app where the work happens.