Does the MCP server have the same tools as the built-in AI assistant?

Yes. The MCP server exposes the exact same 50+ tools that the built-in assistant uses. There is no reduced or simplified toolset. External agents get full access to tables, records, views, workflows, documents, email, files, and search.

How do I authenticate my MCP client?

Create an API key in your Lotics workspace under Settings. Pass it in the x-api-key header when configuring your MCP client. Each key can be named and revoked independently, so you can track which clients are connected.

Which MCP clients are supported?

Any client that supports the MCP protocol over HTTP transport works with Lotics. This includes Claude Code, Claude Desktop, Cursor, Windsurf, Continue, and custom agents built with the MCP SDK. The protocol is standardized, so new clients work automatically.

Is there a rate limit on the MCP server?

The MCP server follows the same rate limits as the Lotics API. For most workspaces, this is generous enough for interactive use and moderate automation. Contact support if you need higher limits for heavy batch processing.

Can I restrict which tools an API key can access?

Currently, an API key grants access to all tools available in the workspace. The key inherits the permissions of the workspace member who created it. We recommend creating dedicated API keys for each use case so you can revoke access granularly.

Lotics MCP Server

Lotics exposes a Model Context Protocol (MCP) server that lets external AI agents interact with your workspace. The same 50+ tools your built-in AI assistant uses, now available to Claude Code, Cursor, Windsurf, and any MCP-compatible client.

What is MCP?

Model Context Protocol (MCP) is an open standard created by Anthropic that defines how AI applications discover and call tools provided by external services. Instead of hardcoding API integrations, an MCP client connects to an MCP server at runtime, receives a list of available tools with their schemas, and invokes them through a standardized interface. This means AI agents can work with any MCP-compatible service without custom integration code.

MCP separates tool discovery from tool execution. The client asks the server "what can you do?" and receives structured descriptions of every available operation: input parameters, types, required fields, and human-readable explanations. The AI model uses these descriptions to decide which tools to call and how to call them, making the integration self-documenting and adaptive.

Because MCP is an open protocol, the ecosystem is growing rapidly. Claude Desktop, Claude Code, Cursor, Windsurf, Continue, and many other AI development tools support MCP natively. Any tool you build for one client works with all of them.

Why Lotics Has an MCP Server

Lotics already has a powerful built-in AI assistant with 50+ tools for managing your operational data. The MCP server exposes the exact same toolset to external AI agents, so you can interact with your Lotics workspace from wherever you work: your IDE, a terminal, a custom automation script, or a chat interface.

This is not a simplified or read-only API wrapper. External agents get full access to create and modify tables, query and update records, configure views, trigger workflows, generate documents, send emails, upload files, and search across your workspace. If the built-in assistant can do it, your external agent can too.

The practical benefit is composability. A Claude Code session can query your inventory table, cross-reference it with shipment records, generate a report, and email it, all in a single conversation. A Cursor agent can read your project tracking data while writing code. Custom agents can run scheduled data pipelines that read from one table and write to another.

Setup and Authentication

Connecting an MCP client to Lotics takes three steps: create an API key, add the server configuration to your client, and start using it. The entire process takes under two minutes.

First, go to Settings in your Lotics workspace and create a new API key. Give it a descriptive name (e.g., "Claude Code" or "CI Pipeline") so you can track usage and revoke it later. The key is shown once. Copy it immediately.

Then configure your MCP client with the server URL and the API key. Lotics uses HTTP transport with streaming responses and authenticates via the x-api-key header. Once configured, tools are auto-discovered at runtime. No manual tool registration needed.

Parameter	Value
Server URL	https://api.lotics.ai/mcp
Transport	HTTP with streaming (streamable-http)
Authentication	x-api-key header
Tool discovery	Automatic at connection time

Available Tools

The MCP server exposes every tool available to the built-in AI assistant. These cover the full surface area of Lotics, from schema management to document generation. Tools are organized by domain and use clear, descriptive names that AI models can reason about effectively.

Each tool includes a detailed schema with parameter descriptions, required/optional fields, and example values. The AI model uses these schemas to construct correct tool calls without needing external documentation.

Category	Operations	Examples
Tables	Create, read, update, delete tables	Create a shipments table, rename a field, delete an unused table
Fields	Add, modify, remove fields with full type configuration	Add a currency field, change a text field to rich text, configure select options
Records	Create, read, update, delete, query with filters and sorts	Add 50 inventory items, find overdue invoices, bulk-update status fields
Views	Create and configure views with filters, sorts, grouping, field visibility	Create a Kanban view grouped by status, set up a filtered view for overdue items
Workflows	Build and manage automated workflows with triggers and steps	Create a workflow that sends an email when a record status changes
Documents	Generate PDFs, fill form templates, create Excel files	Fill a bill of lading template with shipment data, generate an invoice PDF
Email	Send emails with attachments, process incoming email triggers	Email a generated report to the operations team
Files	Upload, download, and manage file attachments on records	Attach a signed contract PDF to a deal record
Search	Full-text search across all tables and records in the workspace	Find all records mentioning a specific customer or PO number

Client Configuration Examples

Below is the configuration for Claude Code, which uses a .mcp.json file in your project root. Other MCP clients follow similar patterns. Consult your client's documentation for the exact format.

For Claude Code, create a .mcp.json file at the root of your project. The server type is "streamable-http" and authentication is passed via the headers object. Once saved, Claude Code will discover all available Lotics tools automatically the next time you start a session.

Client	Config file	Notes
Claude Code	.mcp.json in project root	Use "streamable-http" type with headers for auth
Claude Desktop	claude_desktop_config.json	Add server entry under mcpServers key
Cursor	.cursor/mcp.json	Same format as Claude Code
Windsurf	~/.codeium/windsurf/mcp_config.json	Global config file
Custom agents	Programmatic via MCP SDK	Use @modelcontextprotocol/sdk for TypeScript/Python

Use Cases

Developer workflows are the most immediate use case. A developer using Claude Code or Cursor can ask their AI assistant to query operational data, create records, or trigger workflows without leaving their editor. For example, a developer building a shipping integration can inspect the shipments table schema, read sample records, and test workflows, all from their coding environment.

Automated data pipelines are another strong fit. A scheduled script using the MCP SDK can connect to Lotics, query records that match certain criteria, perform calculations or transformations, and write results back. This is useful for nightly reconciliation, report generation, or syncing data between Lotics and external systems.

Custom AI agents built with the MCP SDK can offer specialized interfaces for specific teams. An operations team might have a Slack bot that queries Lotics for shipment status. A finance team might have a reporting agent that pulls invoice data and generates summaries. Because MCP handles tool discovery automatically, these agents stay up to date as new tools are added to Lotics.