Nexplora

Connect Nexplora to a compatible AI app

Add Nexplora to an app that supports MCP, to access supported Nexplora tools within the scopes you authorize. Add this address, sign in to Nexplora, and choose Allow:

https://nexplora.ai/api/mcp

Two different ways to use MCP

Public discovery at /ai/mcp is a read-only catalog for learning what Nexplora offers. It cannot see your workspace.

Connected-app access at /api/mcp is the private connection on this page. It lets an AI app use the Nexplora workspace you choose, within the permissions and approvals you set.

What you get

  • Ask Nexplora to do anything in plain words, and it does the work in your account.
  • Every app you connected in Nexplora, with all of its actions, plus the actions you built yourself.
  • MY, the apps Nexplora runs for your company: CRM, Support, Analytics, Projects, ERP, Email, Experiments and Social, each with its own tool.
  • Your memory: what Nexplora knows about you and your work, and new things to remember.
  • Your skills and departments for marketing, sales, SEO, finance and more.
  • Your Chat and CodeChat conversations: see them, read them, and pick a chat up where you left off.

You stay in charge: you choose what each app may do when you connect it, and you can disconnect it any time under Connections. Anything that sends, posts, publishes, deploys, spends money or deletes waits for your OK in Nexplora, under Approvals. The app you connected can't approve it for you. If you texted Nexplora recently, you also get a message on that thread.

Use it like it's built in

/nexplora — slash commands

Type / in an app that shows commands: /nexplora, one per MY app (/crm is CRM, /support, /analytics, …), /memory, /continue and /build-tool.

  • /nexplora find the three deals most likely to close this month and draft a nudge for each
  • /crm how many open deals do I have?
  • /memory Acme
  • /continue pick up my last conversation
  • /build-tool post every new deal to my sales channel

@nexplora — mentions

In apps that let you @-mention a connector's resources, bring your Nexplora context into the chat: @nexplora memory, @nexplora apps, @nexplora recent conversations, or one app like @nexplora CRM. They only read, and only what you allowed when you connected.

$nexplora — compatible coding apps

Install the nexplora skill, then type $nexplora in a compatible coding app (or just ask about your apps) and the agent uses your Nexplora connection for apps, memory, conversations and tools. Download SKILL.md

Save the downloaded SKILL.md in the skills directory supported by your coding app.

The skill uses the Nexplora connection below, so connect Nexplora in that app too.

Set it up

AI apps with custom connectors

  1. Open Settings, then Connectors, and choose Add custom connector.
  2. Name it Nexplora and paste https://nexplora.ai/api/mcp.
  3. Choose Connect, sign in to Nexplora, and choose Allow.

On a team or enterprise plan, an owner may need to add the connector first.

AI apps with developer mode

  1. Open Settings, then Apps & Connectors. Under Advanced settings, turn on developer mode.
  2. Choose Create, name it Nexplora, paste https://nexplora.ai/api/mcp, and pick OAuth for sign-in.
  3. Sign in to Nexplora and choose Allow. Then turn Nexplora on in a chat from the tools menu.

Claude Code

Run this, then sign in to Nexplora when asked:

claude mcp add --transport http nexplora https://nexplora.ai/api/mcp

Adding Nexplora from an AI app's connector directory gives you a smaller, listed version (no tool that starts a whole Nexplora task, no tool building). The address on this page is the full connector: add it yourself in Claude, Claude Code, Codex or any other app that speaks MCP when you want your AI to steer Nexplora, build custom tools and run your skills.

Cursor

Add this to ~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{
  "mcpServers": {
    "nexplora": { "url": "https://nexplora.ai/api/mcp" }
  }
}

Open Cursor Settings, then MCP, and choose Log in next to nexplora.

VS Code

Add this to .vscode/mcp.json, then choose Start and sign in when asked:

{
  "servers": {
    "nexplora": { "type": "http", "url": "https://nexplora.ai/api/mcp" }
  }
}

Other MCP apps

Add a remote MCP server with the address https://nexplora.ai/api/mcp. Apps that support sign-in will open Nexplora for you to Allow.

If the app only takes a fixed key, create one in Nexplora under Settings → Developer API with "Connect an AI app over MCP" and only the AI-app permissions it needs, and send it as Authorization: Bearer nxp_…. Treat it like a password.

Bring another server into Nexplora

Add a third-party MCP server

Register the server in All connections, not in Chat. Choose Set it up manually, pick the Tool server tab, then enter a name and the server's HTTPS address.

  1. Choose the sign-in the server uses: sign in, an API key, or no auth. Enter any token in the form's token field. Never paste a token or any other secret into Chat: the conversation is kept, and the model you chose may see it.
  2. Nexplora saves the server. A token goes to your vault, and the connection is marked signed in. Then review the saved connection under MY → Connections and its available access under MY → App permissions.

Once it's connected, Chat can list your connected servers and disconnect one. It can't add one. This is the reverse of the setup above: here Nexplora uses the third-party server; above, another AI app uses Nexplora.

One app at a time (optional)

The address above already reaches all of your apps. If you also want one app's full list of actions as separate tools in your AI app, add that app's own address the same way: https://nexplora.ai/api/mcp/integrations/<app>/mcp

Sign in to see the addresses for the apps you connected.

For developers

  • MCP server: https://nexplora.ai/api/mcp (Streamable HTTP, OAuth 2.1 with PKCE and dynamic client registration). Server card: /.well-known/mcp.json.
  • Prefer REST? The Developer API runs the same actions with the same approval rules: OpenAPI description.
  • MCP uses the versioned JSON-RPC 2.0 protocol over HTTP; the public A2A agent card is available at /.well-known/agent-card.json.