MCP Server¶
Nexplane exposes a Model Context Protocol (MCP) server at /mcp that lets AI agents — Claude, Cursor, or any MCP-compatible client — read infrastructure state, create change requests, and monitor execution results. The server implements the MCP SSE (Server-Sent Events) transport.
Every MCP tool operates within the caller's organization. All proposed changes go through the normal CR lifecycle — the AI proposes, a human approves, and only then does the platform execute. The approval gate is enforced for every AI-assisted action without exception.
Connecting¶
Agent Token¶
All MCP tools require a token parameter — a Nexplane agent token. Tokens are created by an admin via the API:
POST /auth/agent-tokens
Authorization: Bearer <user-jwt>
Content-Type: application/json
{
"name": "claude-desktop",
"scopes": ["read", "write"]
}
The response contains the token value. Store it — it is shown only once.
Admin required
Only users with the admin role can create agent tokens. Token scope can be further restricted by allowed_connector_types, allowed_asset_tags, allowed_cr_types, and allowed_roles.
Claude Desktop¶
Add Nexplane to your claude_desktop_config.json:
{
"mcpServers": {
"nexplane": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-sse", "http://localhost:8000/mcp"]
}
}
}
Claude Code¶
Discovery Pattern¶
Before creating a change request, agents should follow this discovery sequence:
list_connectors— confirm the needed connector is configured and activelist_assets/search_assets/get_asset— resolve the target asset and its contextlist_catalog_actions(connector_type)/list_change_types— find the exact action or CR typecreate_change_request— propose the change
Use planning context tools (get_fleet_context, get_asset_history, get_cross_host_dependency_map) to fill parameters from infrastructure state rather than asking the operator.
Tool Domains¶
| Domain | Tools | Description |
|---|---|---|
| Assets & Connectors | 14 | Enumerate and inspect infrastructure assets and connector status |
| Change Requests | — | Create, approve, execute, and roll back change requests |
| Findings & Identity | — | Manage vulnerability findings and identity access |
| Host Intelligence | — | Live host interrogation via the Nexplane Agent |
| Projects, Planning & Migration | — | Multi-CR project orchestration, migration profiling, reference scanning, runbooks |