FlowDot — Claude Code MCP server: what it does, how to install
FlowDot connects an MCP client to FlowDot workflows, apps, knowledge bases, and agent tools. Use it to build, inspect, and run workflows from Claude Code.
claude mcp add --env FLOWDOT_API_TOKEN=<FLOWDOT_API_TOKEN> --env FLOWDOT_HUB_URL=<FLOWDOT_HUB_URL> --transport stdio flowdot -- npx -y @flowdot.ai/mcp-serverWhat FlowDot does
FlowDot exposes FlowDot capabilities to an MCP client: workflow graphs, custom nodes, apps, knowledge bases, agent toolkits, and community objects. An agent can create workflows, add nodes and connections, validate the graph, start executions, and inspect their status.
The server also covers RAG knowledge bases, agent recipes, sharing, and workflow discovery. Set FLOWDOT_API_TOKEN before connecting it, then run npx -y @flowdot.ai/mcp-server. Recipes are designed through MCP but executed with the FlowDot CLI @flowdot.ai/cli, since they are long-running programs that need local file, code, and shell access.
For Claude Code, the setup uses claude mcp add. The next session loads the server configuration.
What it helps with
- build a workflow graph from nodes
- validate workflow connections and settings
- start a workflow and inspect its status
- design a multi-step agent recipe
- query a knowledge base for RAG
Example prompts for the agent
- Build a FlowDot workflow for processing this data and validate its graph.
- List the available FlowDot nodes and add the needed ones to a new workflow.
- Design an agent recipe with branching and parallel steps.
How to install FlowDot
Platform CLI
claude mcp add --env FLOWDOT_API_TOKEN=<FLOWDOT_API_TOKEN> --env FLOWDOT_HUB_URL=<FLOWDOT_HUB_URL> --transport stdio flowdot -- npx -y @flowdot.ai/mcp-serverReplace <…> with your own values
Platform CLI
codex mcp add flowdot --env FLOWDOT_API_TOKEN=<FLOWDOT_API_TOKEN> --env FLOWDOT_HUB_URL=<FLOWDOT_HUB_URL> -- npx -y @flowdot.ai/mcp-serverReplace <…> with your own values
Config file
{
"mcpServers": {
"flowdot": {
"command": "npx",
"args": [
"-y",
"@flowdot.ai/mcp-server"
],
"env": {
"FLOWDOT_API_TOKEN": "<FLOWDOT_API_TOKEN>",
"FLOWDOT_HUB_URL": "<FLOWDOT_HUB_URL>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"flowdot": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@flowdot.ai/mcp-server"
],
"env": {
"FLOWDOT_API_TOKEN": "<FLOWDOT_API_TOKEN>",
"FLOWDOT_HUB_URL": "<FLOWDOT_HUB_URL>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e FLOWDOT_API_TOKEN=<FLOWDOT_API_TOKEN> -e FLOWDOT_HUB_URL=<FLOWDOT_HUB_URL> flowdot npx -y @flowdot.ai/mcp-serverReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about FlowDot
How do I install FlowDot in Claude Code?
Set FLOWDOT_API_TOKEN, then run claude mcp add mcp-server -e FLOWDOT_API_TOKEN=<FLOWDOT_API_TOKEN> -- npx -y @flowdot.ai/mcp-server. Claude Code loads it in the next session.
What does FlowDot do?
It gives Claude Code MCP access to FlowDot workflows, apps, custom nodes, knowledge bases, and agent toolkits. The client can build, validate, run, and monitor workflows.
Can MCP execute agent recipes?
MCP can design recipes, while execution requires the FlowDot CLI @flowdot.ai/cli.
