Falcon Builder

Official registry Developer tools

Build, debug, and publish production AI agent workflows in Falcon Builder from any MCP client.

Install

Add this to your client's MCP configuration (Claude Desktop: claude_desktop_config.json; Cursor: .cursor/mcp.json; VS Code and others use the same "mcpServers" shape).

Remote (streamable-http)
Claude Code
claude mcp add --transport http falcon-builder 'https://www.falconbuilder.dev/api/mcp' --header 'Authorization: <Falcon Builder API token, created in Settings > Developer. Clients that sign in with OAuth (claude.ai, Claude Desktop, ChatGPT) do not need it.>'
JSON configuration (Claude Desktop, Cursor, VS Code)
{
    "mcpServers": {
        "falcon-builder": {
            "type": "http",
            "url": "https://www.falconbuilder.dev/api/mcp",
            "headers": {
                "Authorization": "<Falcon Builder API token, created in Settings > Developer. Clients that sign in with OAuth (claude.ai, Claude Desktop, ChatGPT) do not need it.>"
            }
        }
    }
}

Documentation

From the project's README on GitHub, refreshed weekly. Links open on GitHub.

Contents
  1. Falcon Builder MCP Server
  2. Connect
  3. Permissions
  4. Tools
  5. Things to ask your assistant
  6. About

Falcon Builder MCP Server

Build, edit, and publish production AI agents from Claude, Cursor, ChatGPT, VS Code, or any MCP client. The Falcon Builder MCP server lets your assistant inspect executions, edit and duplicate workflows, run them, and publish them in your Falcon Builder workspace.

Available on every plan, including Free.

This is a hosted server. Nothing to install or run locally.

Connect

OAuth clients: claude.ai, Claude Desktop, ChatGPT

No token needed. The client registers itself automatically.

claude.ai / Claude Desktop

  1. Settings → Connectors → Add custom connector
  2. Server URL: https://www.falconbuilder.dev/api/mcp
  3. Leave OAuth Client ID and Client Secret empty
  4. Pick a workspace and the permissions to grant, then approve

ChatGPT Settings → Connectors → Create (turn on Developer Mode if prompted). Same URL, OAuth fields left empty.

Token clients: Claude Code, Cursor, VS Code, Gemini CLI

Create a token first: Falcon Builder → Settings → Developer, name it, choose permissions, and copy it. Tokens are shown once. Each workspace can have up to 10 active tokens.

Claude Code

claude mcp add --transport http falcon-builder \
  https://www.falconbuilder.dev/api/mcp \
  --header "Authorization: Bearer YOUR_TOKEN"

Cursor (.cursor/mcp.json)

{
  "mcpServers": {
    "falcon-builder": {
      "url": "https://www.falconbuilder.dev/api/mcp",
      "headers": { "Authorization": "Bearer YOUR_TOKEN" }
    }
  }
}

VS Code (.vscode/mcp.json). VS Code prompts for the token and stores it securely.

{
  "inputs": [
    {
      "type": "promptString",
      "id": "falcon-token",
      "description": "Falcon Builder API token (Settings > Developer)",
      "password": true
    }
  ],
  "servers": {
    "falcon-builder": {
      "type": "http",
      "url": "https://www.falconbuilder.dev/api/mcp",
      "headers": { "Authorization": "Bearer ${input:falcon-token}" }
    }
  }
}

Gemini CLI (~/.gemini/settings.json). Use httpUrl; in Gemini CLI, url means SSE.

{
  "mcpServers": {
    "falcon-builder": {
      "httpUrl": "https://www.falconbuilder.dev/api/mcp",
      "headers": { "Authorization": "Bearer YOUR_TOKEN" }
    }
  }
}

Permissions

You choose these when you connect over OAuth or create a token. Read is always included.

PermissionWhat it allows
ReadInspect workflows, executions, logs, version history, and interfaces
Edit draftsChange workflow drafts and configure interfaces without touching production
Run workflowsExecute workflows with test input. Runs count against your plan limits
Publish & rollbackDeploy drafts to production and restore earlier versions

Tools

Read whoami, switch_workspace, list_agents, list_workflows, get_workflow, get_workflow_definition, get_workflow_node, validate_workflow, list_executions, get_execution, get_execution_logs, list_workflow_versions, list_interfaces, get_interface

Edit drafts create_workflow, edit_workflow, duplicate_workflow, duplicate_agent, create_interface, update_interface, add_interface_access_rule, remove_interface_access_rule, delete_interface

Run workflows run_workflow, rerun_executions

Publish & rollback publish_workflow, rollback_workflow, publish_interface

Every tool declares MCP annotations (read-only, destructive, idempotent, open-world), so clients that honor them can ask before anything changes. run_workflow and rollback_workflow also require an explicit confirm: true.

Things to ask your assistant

  • "Why did the last run of my intake workflow fail? Show me the node that errored."
  • "Duplicate my support agent and change its escalation step to email the on-call address."
  • "Validate the draft of my lead-routing workflow, then publish it."
  • "Roll my onboarding workflow back to the previous version."

About

Falcon Builder is built by NeoSky AI.

Frequently asked questions

What is Falcon Builder?

Build, debug, and publish production AI agent workflows in Falcon Builder from any MCP client.

How do I install Falcon Builder?

It runs as a remote server. In Claude Code: claude mcp add --transport http falcon-builder 'https://www.falconbuilder.dev/api/mcp' --header 'Authorization: <Falcon Builder API token, created in Settings > Developer. Clients that sign in with OAuth (claude.ai, Claude Desktop, ChatGPT) do not need it.>'. For other clients, paste the JSON configuration from this page into their MCP settings.

Does Falcon Builder need an API key?

Yes. It expects Authorization, which you set in its configuration.

Is Falcon Builder open source?

Its code is public on GitHub, but the repository does not declare a standard license.

Who maintains Falcon Builder?

It is published by NeoSky-AI. AI.tn lists it but does not maintain or endorse it.

From the official MCP Registry. Listing is not an endorsement: review a server's code and permissions before giving it access to your data.