After you connect
In your AI app
Once OAuth succeeds, your app sends requests to https://mcp.cortexhub.ai/v1/mcp with your access token. The assistant sees a small set of CortexHub meta-tools for discovering tools, loading their input schemas, invoking them, polling a parked action, and requesting consent. It searches for what it needs and invokes tools through the gateway. When a tool needs your approval, operator review, or a toolkit that is not connected yet, the matching prompt appears in your MCP client.
See How it works for the full flow.
In CortexHub Cloud
Company organization
- Toolkits and limits come from your organisation's enabled catalogue, your groups, and the access rules your admin configured.
- If a toolkit shows connect required, either finish the Connect prompt in your MCP client when the assistant asks, or open My space -> Toolkits in CortexHub Cloud and complete vendor OAuth there. You connect a toolkit once; it is reused across your co-worker and your AI apps.
- If a tool is disabled with a reason, your group may not be entitled to it, or a policy blocked it. Ask your org admin.
- Org admins can see connection status and activity for gateway identities in the organisation. Your own toolkit credentials stay in CortexHub: never in the chat.
Your own organization
- You are the org owner: enable toolkits, configure Groups, set the Policies floor and Security guardrails, and manage members in CortexHub Cloud.
- Each AI app you connect appears under My space -> External agents with its own gateway identity.
In both cases, route external actions through CortexHub rather than pasting secrets into the assistant.
Sessions and activity
Your interactive MCP sessions appear in CortexHub Cloud under My space -> External agents -> Sessions. Org admins see every gateway identity and session for the organisation in Org space. Activity is attributed to your user and the gateway identity for that app.