Creating FlowDot Apps

FlowDot Apps let you create custom React interfaces for your workflows.

App Environment

Apps run in a sandboxed browser with:

• React 18 (global - use React.useState, etc.)
• Tailwind CSS (full utility classes)
• FlowDot color tokens (primary-50 to primary-900)

Code Rules

1. No imports - React is global
2. No exports - Just define your function
3. Function must be named: function MyAppName() { ... }
4. Use Tailwind CSS for all styling

Basic App Structure

function MyApp() {
  const [result, setResult] = React.useState(null);
  const [loading, setLoading] = React.useState(false);
  
  const handleSubmit = async () => {
    setLoading(true);
    const response = await invokeWorkflow('workflow-hash', {
      inputName: 'value'
    });
    setResult(response);
    setLoading(false);
  };
  
  return (
    <div className="p-4">
      <button 
        onClick={handleSubmit}
        className="bg-primary-600 text-white px-4 py-2 rounded"
      >
        {loading ? 'Running...' : 'Run Workflow'}
      </button>
      {result && <pre>{JSON.stringify(result, null, 2)}</pre>}
    </div>
  );
}

Linking Workflows

1. Create your app
2. Go to app settings
3. Link workflows by hash
4. Use invokeWorkflow('hash', inputs) in your code

Display Modes

• Windowed: Standard view with FlowDot header
• Fullscreen: Full viewport, minimal controls
• Embedded: No FlowDot UI, for iframe embedding

Multi-File Apps

Organize code across multiple files:

• Component files (.jsx, .tsx)
• Utility files (.js, .ts)
• Style files (.css)
• JSON configuration

FlowDot API Overview

The FlowDot API provides programmatic access to all platform features.

Base URLs

• Hub API: https://flowdot.ai/api
• Node Server: Proxied through /api/node-core-api

Authentication

All API requests require authentication via Bearer token:

curl -H "Authorization: Bearer YOUR_TOKEN" \
  https://flowdot.ai/api/hub/workflows

Token Types

Token Type	Prefix	Use Case
Sanctum Token	None	Web session authentication
MCP Token	fd_mcp_	External integrations (Claude, Cursor)

Response Format

All responses are JSON:

{
  "success": true,
  "data": { ... },
  "message": "Optional message"
}

Error Responses

{
  "success": false,
  "error": "ERROR_CODE",
  "message": "Human-readable error description",
  "errors": { ... }  // Validation errors (422)
}

Rate Limiting

Rate limits vary by subscription tier:

Tier	Requests/Minute
Free	30
Creator	60
Pro	120
Team	300

Exceeding limits returns HTTP 429.

Identifiers

FlowDot uses different ID formats:

• Workflows, Apps, Custom Nodes: 10-character hash (e.g., aB3xY9kL2m)
• Executions, Teams, Schedules: Numeric IDs
• Presets, Shared Results: 8-character hash

Two-Factor Authentication

Add an extra layer of security to your FlowDot account with TOTP-based 2FA.

Setting Up 2FA

1. Go to Settings → Security
2. Click Enable Two-Factor Authentication
3. Scan the QR code with your authenticator app:
   • Google Authenticator
   • Authy
   • 1Password
   • Any TOTP-compatible app
4. Enter the 6-digit code from your app
5. Save your recovery codes

Recovery Codes

When you enable 2FA, you receive 8 recovery codes:

• Each code can only be used once
• Store them securely (password manager, printed copy)
• Use a recovery code if you lose access to your authenticator
• You can regenerate codes (invalidates old ones)

Logging In with 2FA

1. Enter your email and password
2. Enter the 6-digit code from your authenticator
3. Or use a recovery code

Disabling 2FA

1. Go to Settings → Security
2. Click Disable Two-Factor Authentication
3. Enter your password to confirm

Security Tips

• Use a password manager with TOTP support
• Keep recovery codes in a separate secure location
• Don't share your 2FA codes with anyone
• Regenerate recovery codes if you suspect compromise

Team Management

Teams let you collaborate on workflows, documents, and resources.

Creating a Team

1. Go to Teams page
2. Click Create Team
3. Enter team name and description
4. You become the team owner

Team Roles

Role	Permissions
Owner	Full access, can delete team
Admin	Manage members, full resource access
Manager	Edit resources, limited member management
Editor	Edit shared resources
Member	View and use shared resources

Inviting Members

1. Go to your team settings
2. Click Invite Member
3. Enter their email address
4. Select their role
5. They'll receive an invitation email

Shared Resources

Teams can share:

• Workflows: Team-visible workflows
• Documents: Team knowledge base
• Categories: Shared document categories

Transferring Ownership

To transfer team ownership:

1. Go to team settings
2. Click Transfer Ownership
3. Select the new owner (must be admin)
4. Confirm the transfer

Creating Schedules

Schedules let you run workflows automatically at specified times.

Schedule Types

Cron Schedules

Use cron expressions for precise timing.

Common patterns:

• 0 9 * * * - Every day at 9 AM
• 0 */2 * * * - Every 2 hours
• 0 9 * * 1-5 - Weekdays at 9 AM
• 0 0 1 * * - First day of each month

Interval Schedules

Simple minute-based intervals.

• Every 15 minutes
• Every hour
• Every 6 hours

Creating a Schedule

1. Go to Schedules page
2. Click Create Schedule
3. Select the workflow to run
4. Choose schedule type (Cron or Interval)
5. Configure the timing
6. Enable the schedule

Managing Schedules

• Pause/Resume: Toggle schedule without deleting
• Run Now: Execute immediately
• View History: See past executions

Schedule Execution Logs

Each scheduled run is logged with:

• Execution timestamp
• Status (success/failure)
• Duration
• Error details (if any)

Tips

• Test your workflow manually before scheduling
• Start with longer intervals and decrease as needed
• Check execution history regularly
• Use meaningful schedule names

Setting Up Integrations

Integrations let external services trigger your workflows via webhooks.

Supported Services

Service	Event Types
Discord	Message, reaction, member join
Slack	Message, reaction, slash command
GitHub	Push, PR, issue, star
Microsoft Teams	Message, reaction
Jira	Issue created/updated
Trello	Card, list, board events
Figma	File update, comment

Creating an Integration

1. Go to Integrations page
2. Click Add Integration
3. Select the service
4. Follow the setup instructions
5. Configure webhook URL in the external service

Event Mapping

Map events to workflows:

1. In your integration settings, click Add Event Mapping
2. Select the event type (e.g., "message", "push")
3. Choose the target workflow
4. Optionally add filter conditions
5. Enable the mapping

Webhook Security

All webhooks use signature verification:

• Discord: Ed25519 signatures
• Slack: HMAC-SHA256
• GitHub: HMAC-SHA256

Invalid signatures are rejected automatically.

Using Webhook Data in Workflows

The webhook payload is available as workflow input:

{
  "vendor": "github",
  "event_type": "push",
  "payload": {
    "ref": "refs/heads/main",
    "commits": [...]
  }
}

Setting Up Knowledge Base

The Knowledge Base lets you upload documents and use semantic search (RAG) in your workflows.

Supported Formats

• PDF documents
• Word documents (.docx)
• Text files (.txt)
• Markdown files (.md)
• CSV files
• JSON files

Uploading Documents

1. Go to Knowledge Base page
2. Click Upload Document
3. Choose a file or enter a URL
4. Select a category (optional)
5. Wait for processing to complete

Document Processing

When you upload a document:

1. Content is extracted from the file
2. Text is split into chunks
3. Each chunk gets an embedding (vector representation)
4. Chunks are indexed for semantic search

Using RAG in Workflows

1. Add Knowledge Base Node

Configure which categories to search.

2. Add RAG Query Node

Connect a text input to the query.

RAG Query returns:

• Relevant document chunks
• Source attribution
• Confidence scores

3. Connect to LLM Node

Send RAG results to an LLM for context-aware responses.

Categories

Organize documents into categories:

• Personal categories (your own)
• Team categories (shared with team)

Team Knowledge Bases

Share documents with your team:

1. Create a team category
2. Upload documents to that category
3. Team members can search these documents

CLI Authentication Guide

The FlowDot CLI uses a secure device authorization flow that doesn't require you to enter your password in the terminal.

How It Works

1. Initiate: CLI requests an auth token from the server
2. Browser: User visits the auth URL and logs in
3. Confirm: User confirms the device authorization
4. Complete: CLI receives an API token

Implementation

Step 1: Initiate the Flow

const response = await fetch('https://flowdot.ai/api/cli/auth/initiate', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({ device_name: 'My CLI App' })
});

const { auth_token, auth_url } = response.data;
console.log(`Please visit: ${auth_url}`);

Step 2: Poll for Completion

async function pollForToken(authToken) {
  while (true) {
    const response = await fetch(
      `https://flowdot.ai/api/cli/auth/verify/${authToken}`
    );
    const { data } = await response.json();

    if (data.status === 'confirmed') {
      return data.api_token;
    }
    if (data.status === 'expired') {
      throw new Error('Auth session expired');
    }

    await new Promise(r => setTimeout(r, 2000));
  }
}

Step 3: Store Token Securely

Use the system keychain (not plain text files):

import keytar from 'keytar';

await keytar.setPassword('flowdot', 'api-token', token);

Security Notes

• Auth tokens expire after 15 minutes
• API tokens should be stored in the system keychain
• Never log or display the full API token

Creating Agent Toolkits

Agent Toolkits are collections of tools that extend the AI Agent's capabilities.

Toolkit Structure

A toolkit contains:

• Tools: Individual actions (HTTP calls or workflow triggers)
• Credential Requirements: What API keys are needed
• Metadata: Name, description, category, tags

Tool Types

HTTP Tools

Make REST API calls to external services.

{
  "name": "get-weather",
  "title": "Get Weather",
  "tool_type": "http",
  "endpoint_config": {
    "url": "https://api.weather.com/v1/current",
    "method": "GET",
    "query_params": {
      "key": "{{credential.API_KEY}}",
      "q": "{{input.city}}"
    }
  }
}

Workflow Tools

Execute a FlowDot workflow as a tool.

Creating a Toolkit

1. Go to Agent Toolkits page
2. Click Create Toolkit
3. Enter name, title, and description
4. Add tools to the toolkit
5. Define credential requirements
6. Test and publish

Credential Mapping

Toolkits use placeholders for credentials:

• {{credential.API_KEY}} - References a required credential
• {{input.parameter}} - References a tool input

When users install your toolkit, they map their own API keys to these placeholders.

Installing Toolkits

1. Browse public toolkits or get a direct link
2. Click Install
3. Map required credentials to your saved API keys
4. Tools become available in the Agent

Using the AI Agent

The AI Agent provides a conversational interface for executing workflows and performing research.

Getting Started

1. Navigate to /agent
2. Start typing your request
3. The Agent can execute workflows, search the web, and use installed toolkits

Enabling Workflows for Agent

To make a workflow available to the Agent:

1. Open the workflow in the Editor
2. Click the Star icon to favorite it
3. Favorited workflows appear in the Agent's available tools

Agent Capabilities

Workflow Execution

Ask the Agent to run your workflows naturally:

• "Generate a summary of this article"
• "Create an image with my workflow"
• "Run the data analysis workflow"

Built-in Research Tools

The Agent has web research capabilities:

• Web Search: Search the internet for information
• Fetch URL: Read content from any webpage

Toolkit Integration

Use installed toolkit tools in conversations:

• API integrations
• Custom HTTP tools
• Workflow-based tools

Conversation Features

• Image Attachments: Upload up to 5 images (10MB each)
• Context Memory: Agent remembers conversation history
• Token Compression: Long conversations automatically compress

Voice Mode

For hands-free operation:

1. Click the Voice button
2. Speak your request
3. Agent responds with audio
4. Real-time conversation with interruption support

Creating Custom Nodes

Custom nodes let you extend FlowDot with your own JavaScript logic.

Custom Node Structure

Every custom node requires a processData function:

function processData(inputs, properties, llm) {
  // Access inputs by name
  const myInput = inputs.InputName || '';
  
  // Access configured properties
  const myProp = properties.propertyKey;
  
  // Your logic here
  const result = myInput.toUpperCase();
  
  // Return outputs (keys must match output names)
  return {
    OutputName: result
  };
}

Data Types

Supported input/output data types:

• text - String values
• number - Numeric values
• boolean - True/false
• json - JSON objects
• array - Arrays
• any - Any type

Adding LLM Capability

Enable llm_enabled to call AI from your custom node:

function processData(inputs, properties, llm) {
  const result = llm.call({
    prompt: `Summarize: ${inputs.Text}`,
    temperature: 0.5
  });
  
  return {
    Summary: result.success ? result.response : result.error
  };
}

Execution Limits

• Timeout: 1-30 seconds (default: 5s)
• Memory: 32-256 MB (default: 128 MB)

Publishing Custom Nodes

Share your custom nodes with the community:

1. Create and test your node
2. Go to Custom Nodes page
3. Click your node's settings
4. Set visibility to Public
5. Others can now copy and use your node

Using LLM Nodes

LLM (Large Language Model) nodes let you integrate AI into your workflows.

Available LLM Nodes

LLM Query Node

The main node for AI text generation.

Inputs:

• Prompt: The text to send to the AI
• System Prompt (optional): Instructions for the AI's behavior

Outputs:

• Response: The AI's generated text

LLM JSON Node

Returns structured JSON output from the AI.

LLM Thinking Node

Chain-of-thought reasoning for complex problems.

Quick Select Model Tiers

Use the Quick Select buttons to choose model complexity:

Tier	Description	Best For
FlowDot	Platform credits, no API key needed	Getting started
Simple	Fast, basic models	Straightforward tasks
Capable	Mid-level performance	Most use cases
Complex	Advanced reasoning	Complex problems

Configuring Temperature

Temperature controls creativity:

• 0.0: Focused, deterministic responses
• 0.7: Balanced (default)
• 1.0: Creative, varied responses

Using Your Own API Keys (BYOK)

To use your own API keys:

1. Go to Settings → LLM BYOK
2. Add your provider API key (OpenAI, Anthropic, etc.)
3. Go to Settings → Preferred Models
4. Configure which provider/model to use for each tier

Local Models with Ollama

Run AI locally without cloud APIs:

1. Install Ollama on your machine
2. In FlowDot, enable Ollama Local Browser provider
3. Select local models in your LLM nodes

Building Your First Workflow

Workflows are visual automation pipelines that connect nodes to process data.

Understanding Workflows

A workflow consists of:

• Nodes: Processing units that perform actions
• Connections: Data flow between nodes
• Inputs: User-provided data when running
• Outputs: Results displayed after execution

Node Categories

FlowDot includes 14 node categories:

1. Input: Text, Image, Audio, File Upload, Toggle
2. AI & Language: LLM Query, Vision, Text-to-Speech, Speech-to-Text
3. Output: Text, Audio, PDF, PowerPoint
4. Flow Control: Conditional Branch, Decision Router, Loops, Delay
5. Data Processing: JSON Assembly, List Manipulation, Data Mapping
6. Web & APIs: HTTP Request, Web Browser, Internet Search
7. Image Processing: Image Manipulation
8. File System: File operations
9. Knowledge Base: RAG Query, Knowledge Base Config
10. Integrations: Discord, Slack, Twitter, LinkedIn, and more
11. Developer Tools: Script Node, Benchmark
12. Variables: Set/Get workflow variables
13. Utilities: Data Mapper, Text Manipulation
14. Visualization: Data charts and graphs

Connecting Nodes

1. Click and drag from an output socket (right side of node)
2. Drop onto an input socket (left side of target node)
3. Compatible socket types will highlight
4. Delete connections by clicking on them

Saving Workflows

1. Click the Settings button in the toolbar
2. Enter a name and optional description
3. Choose visibility (Private, Public, Team, Unlisted)
4. Click Save

Running Workflows

From the Editor:

1. Click Run Workflow
2. Fill in any required inputs
3. Watch real-time execution

From the Dashboard:

1. Navigate to /workflow/{hash}
2. Enter inputs in the form
3. Click Run
4. View streaming results

Quick Start Guide

Create your first FlowDot workflow in just 5 minutes!

Prerequisites

• A FlowDot account (free tier works)
• No API keys required (uses FlowDot Internal credits)

Step 1: Create a New Workflow

1. Navigate to the Workflows page
2. Click Create New Workflow
3. You'll be taken to the visual Editor

Step 2: Understand the Editor

The workflow editor has three main areas:

• Node Palette (left): Categorized list of available nodes
• Canvas (center): Drag-drop workspace for building
• Properties Panel (right): Configure selected node settings

Step 3: Add Your First Nodes

Drag these three nodes onto the canvas:

1. Text Input Node (Input category)

• Collects user input when the workflow runs
• Configure the label (e.g., "Enter your question")

2. LLM Query Node (AI & Language category)

• Sends text to an AI model for processing
• Use Quick Select buttons to choose model tier
• Default uses FlowDot Internal (no API key needed)

3. Text Output Node (Output category)

• Displays the result when workflow completes

Step 4: Connect the Nodes

1. Text Input → LLM Query: Connect the Text output to the Prompt input
2. LLM Query → Text Output: Connect the Response output to the Text input

Tip: Click and drag from output sockets (right side) to input sockets (left side).

Step 5: Save and Run

1. Click the Settings button (gear icon)
2. Enter a workflow name
3. Click Save
4. Click Run Workflow
5. Enter your question and click Run

What's Next?

• Add branching logic with Conditional Branch nodes
• Use your own API keys in Settings → LLM BYOK
• Enable the workflow for the Agent by favoriting it
• Share publicly via Workflow Settings → Make Public

Workflow Execution Modes

FlowDot supports two execution modes for different use cases.

Synchronous Execution (Default)

• Executes immediately
• Blocks until complete
• Returns results directly
• Best for quick workflows (under 30 seconds)

When to use:

• Interactive workflows with immediate feedback
• Short processing tasks
• Real-time chat applications

Asynchronous Execution

• Returns an execution ID immediately
• Processes in the background
• Poll for status to check completion
• Best for long-running workflows

When to use:

• Workflows that take more than 30 seconds
• Batch processing tasks
• Workflows with external API calls that may timeout

Choosing Execution Mode

In Workflow Settings, you can set the default execution mode:

1. Sync - Default, immediate execution
2. Async - Background execution with polling

Monitoring Async Executions

The Active Executions Panel shows:

• Currently running background executions
• Execution ID and status
• Start time
• "Watch Live" button to follow progress

Platform Overview

FlowDot provides multiple interfaces for different use cases.

Primary Interfaces

1. Editor (/editor)

The visual workflow builder where you create and edit workflows.

Features:

• Drag-drop node canvas
• Node palette with 14 categories
• Properties panel for configuration
• Real-time connection drawing
• Save and version workflows

2. Dashboard/Runner (/workflow/{hash})

Execute workflows and view results.

Features:

• Dynamic input collection
• Real-time execution streaming
• Result visualization
• Execution history
• Share results publicly

3. Agent (/agent)

Conversational AI interface for natural language workflow execution.

Features:

• Chat-style interaction
• Execute favorited workflows via conversation
• Use installed toolkit tools
• Voice mode for hands-free operation
• Built-in web research capabilities

Secondary Interfaces

Apps (/apps)

Create React frontends powered by your workflows.

Custom Nodes (/custom-nodes)

Build reusable nodes with JavaScript.

Integrations (/integrations)

Connect external services like Discord, Slack, and GitHub.

Schedules (/schedules)

Automate workflow execution on a schedule.

Teams (/teams)

Collaborate with team members on shared resources.

Managing API Keys

FlowDot supports Bring Your Own Keys (BYOK) for LLM providers and MCP tokens for external integrations.

LLM API Keys

Use your own API keys from AI providers:

1. Go to Settings → LLM BYOK
2. Click Add Key
3. Select the provider (OpenAI, Anthropic, etc.)
4. Enter your API key
5. Save

Supported Providers:

• OpenAI
• Anthropic
• Google (Gemini)
• Mistral
• Cohere
• And more...

Preferred Models

Configure which model to use for each tier:

1. Go to Settings → Preferred Models
2. For each tier (Simple, Capable, Complex):
   • Select the provider
   • Select the model
3. Save

Your preferences apply across all workflows.

MCP Tokens

MCP tokens let AI assistants (Claude, Cursor, etc.) access your FlowDot resources.

Creating an MCP Token

1. Go to Settings → MCP Tokens
2. Click Create Token
3. Enter a name (e.g., "Claude Desktop")
4. Select scopes (permissions)
5. Copy the token (shown only once)

Token Scopes

Grant specific permissions:

• workflows:read - View workflows
• workflows:execute - Run workflows
• agent:chat - Use the Agent
• apps:read - View apps
• And more...

Security Best Practices

• Only grant necessary scopes
• Rotate tokens periodically
• Delete unused tokens
• Never share tokens publicly

Publishing and Sharing Workflows

Share your workflows with the community or create shareable execution results.

Workflow Visibility

Visibility	Who Can Access
Private	Only you
Public	Anyone (listed in browser)
Team	Team members only
Unlisted	Anyone with direct link

Making a Workflow Public

1. Open the workflow in the Editor
2. Click Settings (gear icon)
3. Set visibility to Public
4. Save

Public workflows appear in the Community Browser where others can:

• View the workflow structure
• Copy it to their account
• Vote and comment

Sharing Execution Results

Share specific workflow outputs:

1. Run your workflow
2. Click Share Results
3. Add a title and description
4. Set expiration (optional)
5. Copy the shareable link

Recipients can view results without a FlowDot account.

Community Input Presets

Let users share pre-configured inputs:

1. Enable Community Inputs in workflow settings
2. Users can save and share input configurations
3. Others can load these presets with one click

Workflow Provenance

FlowDot tracks fork history:

• See who created the original workflow
• View the full ancestry chain
• Track how many times your workflow was copied

Agent Toolkits are collections of tools that extend the AI Agent's capabilities. They can contain HTTP tools (call external REST APIs) and Workflow tools (trigger your FlowDot workflows). Install toolkits to give the agent new skills.

Browse the Marketplace in Agent Toolkits, find a toolkit you want, click Install, map the required credentials to your stored API keys, and complete installation. The agent can then use those tools automatically.

Yes! Go to My Toolkits, click Create New Toolkit, add tools (HTTP or Workflow type), define credential requirements, and save. You can keep it private or publish it for others.

The agent reads tool descriptions to understand when to use each tool. Write clear descriptions like "Get current weather for a city" so the agent knows to use it for weather-related questions.

Yes. Your actual API keys are never exposed to the toolkit. The system substitutes values at runtime, and keys are stored encrypted in your account. Toolkit authors never see your credentials.

Install the MCP server with npm install -g @flowdot.ai/mcp-server, then configure it in your Claude Desktop or Cursor settings. This provides 80+ tools for full FlowDot access directly from these AI assistants. You can create workflows, manage knowledge bases, and execute automations without leaving your preferred AI interface.

MCP tokens have 30+ granular scopes for fine-grained permission control. Create tokens in User Settings > API Keys > MCP Tokens. Assign only the scopes needed - for example, "workflows:read" and "workflows:execute" for read-only workflow access. Tokens can be revoked anytime, and you can create multiple tokens for different applications.

The AI Agent is a conversational interface that lets you interact with your workflows using natural language. Describe what you want to accomplish, and the agent analyzes your request, selects appropriate workflows, gathers required inputs, and executes them.

In Workflow Settings, toggle "Enable for Agent". The agent uses your workflow's name, description, and tags to understand when to use it, so write these clearly.

The Agent maintains conversation context across multiple interactions, can run multiple workflows in sequence based on your goals, asks clarifying questions when needed, and adapts based on your responses.

Yes! Install Agent Toolkits to give the agent access to external APIs and services. When you chat with the agent, it automatically decides which tools to use based on your request.

The Agent is available on the web platform, through the FlowDot CLI, and via external integrations like Discord, Slack, and Microsoft Teams when configured.

Micro-conversation mode provides condensed responses optimized for edge devices and voice interfaces. Responses are brief, TTS is pre-generated for instant playback, and formatting is simplified. Enable it in Agent settings when using mobile devices, smart speakers, or when you prefer concise interactions.

The Agent automatically extracts and remembers preferences, facts, and patterns from conversations. Memories are shown with toast notifications for transparency. You can edit, disable, or delete specific memories. Over time, the Agent learns your workflow preferences, common tasks, and communication style for increasingly personalized assistance.

Visit the platform, click Sign Up, and register with email/password or use Google/GitHub OAuth for quick signup.

Yes! OAuth login is supported for both Google and GitHub accounts for convenient single sign-on.

Go to Security Settings, click Enable 2FA, scan the QR code with an authenticator app (like Google Authenticator or Authy), enter the verification code, and save your recovery codes in a safe place.

Use the "Forgot Password" link on the login page to receive a reset email.

Go to account settings and look for the delete account option. This action is permanent, cannot be undone, and removes all data per GDPR compliance.

Install the FlowDot CLI with npm install -g @flowdot.ai/cli. After installation, authenticate with flowdot auth login and start using it with natural language commands like flowdot "help me analyze this codebase". The CLI provides local code tools, recipe execution, and integration with the FlowDot platform.

Recipes are multi-step agentic orchestrations that run locally on your machine via the CLI, while Workflows run on FlowDot servers. Recipes support human approval gates, complex branching, loops, and have direct file system access. They're designed for long-running tasks that need local execution and human oversight. Link recipes with flowdot recipes link <hash> <alias> and run them with flowdot my-alias "task".

The COMMS relay enables bidirectional communication between your CLI and Telegram/Discord. Start a relay with /relay telegram in a CLI chat session. This allows you to approve CLI permissions from your phone, receive outputs on mobile, and send commands remotely. It's perfect for monitoring long-running tasks or approving sensitive operations while away from your computer.

Use flowdot loop <interval> "task" to create recurring tasks. Intervals can be minutes (5m), hours (2h), days (1d), or cron expressions. Manage loops with flowdot loops list/pause/resume/stop. Loops run as a background daemon that survives terminal closure, with a default 3-day expiration for safety. View history with flowdot loops history <id>.

The CLI uses a 5-option permission system for sensitive operations: Yes-now (single), Yes-session (until restart), Yes-forever (persistent), No-now (single denial), and No-banned (permanent). Permissions are category-based (command-execute, file-read, file-write) and can be managed with flowdot permissions. This ensures you maintain full control over what the AI can do on your system.

Daemon mode (flowdot agent --listen) runs the CLI as a background service that listens for commands from Telegram/Discord. This enables remote recipe execution with local file access, making it perfect for automation tasks, server management, or running recipes triggered by external events. The daemon maintains a heartbeat with the hub for reliability.

Open your workflow in the editor, go to Settings (gear icon), toggle "Make workflow public", and save. Your workflow appears in the Public Workflow Browser for everyone to discover.

Input presets are saved configurations of workflow inputs that you can share with others. They're useful for creating test cases, sharing common configurations, and documenting how to use workflows.

After running a workflow, click "Share Results" to create a public link. Add an optional title and description, set expiration (never to 1 year), and copy the URL. Anyone with the link can view the results without running the workflow.

Upvote (thumbs up) adds +1, downvote (thumbs down) adds -1 to content scores. Click the same button again to remove your vote. You can vote on public workflows, comments, shared results, and input presets. You must be logged in to vote.

In Workflow Settings, you can see the copy count showing how many times your workflow has been copied. Original creators get credit, and copy chains track derivatives.

Custom nodes let you extend FlowDot with your own logic written in JavaScript. You define inputs, outputs, and a processData function that transforms data however you need.

Go to Custom Nodes, click Create New, define inputs and outputs with their data types (text, number, boolean, json, array, any), write your JavaScript processData function, and save. The node becomes available in your workflow palette.

Yes! Enable LLM capability when creating the node. Users will see model selection buttons, and your script can use llm.call({ prompt, systemPrompt, temperature, maxTokens }) to make AI requests.

Yes! Set your custom node visibility to Public to list it in the marketplace. Others can search, install, and use your nodes. You can also keep nodes Private or Unlisted (accessible via direct link only).

FlowDot tracks energy consumption and CO2 emissions for every AI model execution. Visit /environmental-impact to see detailed metrics. Each model shows energy usage in kWh and estimated CO2 emissions based on data center locations. This transparency helps you make eco-conscious choices when selecting models for your workflows.

Generally, smaller models like GPT-4o-mini and Claude Haiku are most energy-efficient. Local models via Ollama have zero cloud emissions. The model browser shows efficiency ratings - look for the leaf icon for eco-friendly options. Consider using the Simple tier for routine tasks to minimize environmental impact while maintaining effectiveness.

While FlowDot doesn't directly offer carbon offsets, we provide detailed emissions data so you can make informed choices. You can minimize impact by using efficient models, batching operations, and leveraging local AI when possible. The platform helps you track cumulative emissions to understand your AI carbon footprint.

FlowDot Apps are custom React applications you can build that use your workflows as backends. They let you create polished user interfaces for your automations without separate hosting.

Basic React knowledge helps, but FlowDot Apps use a simplified environment with React 18 available globally (just use React.useState, etc.) and Tailwind CSS for styling. No imports or exports needed.

Use the link_app_workflow function to connect workflows to your app. Then call invokeWorkflow('workflow-hash', { inputs }) from your app code to execute workflows and get results.

Yes! Set your app to public visibility to list it in the FlowDot app marketplace. Others can discover, use, and clone your apps.

Three modes: Windowed (standard view with FlowDot header), Fullscreen (full viewport with minimal floating controls), and Embedded (no FlowDot UI, for iframe embedding).

FlowDot is a visual workflow automation platform that lets you create AI-powered automations without coding. You can build workflows using drag-and-drop nodes, execute them through multiple interfaces (web, mobile, CLI), and integrate with popular services like Discord, Slack, GitHub, and more.

Workflows are visual sequences of connected actions (nodes) that automate tasks. They help by combining multiple AI models and services to accomplish complex tasks automatically - from data processing to content generation to app integrations.

No coding required! FlowDot's visual editor uses drag-and-drop nodes. Advanced users can create custom nodes with JavaScript or use the API, but the core platform is designed for no-code automation.

Editor: Create and modify workflows visually with the drag-and-drop canvas (web only). Workflow Runner: Execute finished workflows with a streamlined interface showing inputs and outputs. Agent: Run multiple workflows conversationally using natural language.

FlowDot offers four tiers: Free ($0 - 5 executions/day, 5K tokens/day), Creator ($9/month - 1,000 executions/month, 250K tokens/hour), Pro ($29/month - 5,000 executions/month, 5M tokens/month), and Business ($100/month - 25,000 executions/month, 20M tokens/month). Yearly billing saves ~17-20%. Details at https://flowdot.ai/subscriptions

Yes! The Free tier includes 5 daily workflow executions, 5,000 FlowDot Internal tokens per day, the visual workflow editor, agent interface, public workflow browsing, apps access, and the ability to use your own API keys for unlimited LLM usage. Premium features like scheduling, integrations, teams, and analytics unlock at the Creator tier ($9/month).

A modern web browser with JavaScript enabled (Chrome, Firefox, Safari, Edge), a stable internet connection, and 4GB RAM minimum. The browser must support WebSocket and FileAPI.

Sign up, navigate to the Editor, drag nodes from the palette onto the canvas, connect them by clicking output ports and dragging to input ports, configure each node's settings in the Properties Panel, then save and run your workflow.

Yes! Browse the Public Workflow Browser to find community-shared workflows. You can search by keyword, filter by AI-enabled workflows, sort by votes/comments/favorites, and copy any public workflow to customize for your needs.

Content generation, data processing, file transformations, API integrations, social media automation, document analysis with RAG, image generation, custom React app creation, chatbot responses, scheduled reports, and much more.

Check the documentation at https://flowdot.ai/learning-center, browse community workflows for examples, and use the AI assistant in the editor for guidance.

Chat/Messaging (Discord, Slack, Microsoft Teams), Development (GitHub, Figma, Jira), Social Media (Twitter/X, LinkedIn), and any REST API via HTTP nodes and webhooks.

Inbound: External services trigger your workflows (e.g., Discord message runs a workflow). Outbound: Workflows send data to external services (e.g., workflow posts to Slack).

Create an application in Discord Developer Portal, create a bot, copy the Bot Token/Application ID/Public Key, invite the bot to your server, then enter the credentials in FlowDot's Integration settings.

For event-driven integrations like GitHub, you can map specific events to specific workflows. For example, map "Push to main branch" to trigger your deployment workflow automatically.

Yes! Use HTTP/webhook nodes to connect any API-based service. Configure the endpoint URL, HTTP method, headers, and body to call any REST API.

The Knowledge Base lets you upload documents (PDF, Word, TXT, Markdown, CSV, JSON up to 50MB) and search them using AI-powered semantic search. Use it to give your workflows and the AI agent context from your own documents.

RAG (Retrieval Augmented Generation) lets AI use your documents as context. Instead of relying only on its training, the AI searches your Knowledge Base for relevant content and uses it to generate more accurate, grounded responses.

Navigate to Profile > Knowledge Base, click Upload Documents, drag and drop files or browse to select them, optionally assign categories, and click Upload. Documents are automatically processed and become searchable.

Usually 10-60 seconds depending on document size. Documents go through Pending > Processing > Ready states. Large PDFs may take longer. The status updates automatically.

Add a RAG Query node to your workflow. Configure which category to search, connect a search query input, and connect the output to an LLM node. The AI receives relevant document chunks as context for generating responses.

Keyword search matches exact words. Semantic search understands meaning - searching for "company earnings" can find content about "revenue" and "profit" even without those exact words. The Knowledge Base uses semantic search.

Storage limits vary by subscription plan. Higher tiers (Creator, Pro, Business) have increased storage allowances. Check your current storage usage in the Knowledge Base section of your profile.

FlowDot Internal (free built-in tokens), OpenAI (GPT-4 Turbo, GPT-4, GPT-3.5 Turbo), Anthropic (Claude 3 Opus, Sonnet, Haiku), Google AI (Gemini 1.5 Pro, Gemini Pro), OpenRouter, and local models via Ollama.

No! FlowDot Internal provides free tokens to get started. Add your own API keys when you need more capacity, specific models, or want to use your own billing.

Click the hamburger menu (top right), select API Settings, go to the LLM API Keys tab, enter your API key for each provider, click Test API Key to verify, then Save Settings.

Configure preferred models for different complexity tiers (FlowDot, Simple, Capable, Complex) in API Settings. Then use tier buttons in LLM nodes for instant model selection instead of manually choosing provider/model each time.

Yes! Install Ollama (https://ollama.ai), download models, and connect in FlowDot by selecting Ollama (Local - Browser) in LLM nodes. Your data stays on your machine for privacy.

Yes. All keys are encrypted in the database, user-specific (never shared), never exposed in workflow definitions, and never visible in execution logs. Only you can see your own keys.

No! When you use your own API keys (OpenAI, Anthropic, etc.), those don't count against your FlowDot token allocation. You're billed directly by the provider.

FlowDot supports 7+ search providers: DuckDuckGo (free, default), Google, Bing, Brave, SerpAPI, Serper, and Reddit. Configure provider priority in settings. The system uses hybrid execution - trying client-side first, then server fallback on CORS issues. DuckDuckGo works without API keys, while others require BYOK setup.

When using the Agent or Recipe system, the AI autonomously chains multiple searches and web fetches to gather comprehensive information. It intelligently decides when to search deeper, fetch specific pages, or combine multiple sources. This creates research workflows that adapt based on findings rather than following rigid patterns.

Yes, you can use domain filtering in search nodes. Include specific domains with site: operators or configure blocked domains in settings. This is useful for searching documentation sites, academic sources, or excluding unreliable sources. Reddit search is available as a dedicated provider for community insights.

Navigate to Schedules, click Add Schedule, select your workflow, configure timing using cron expressions or simple intervals, set your timezone, and save. Workflows run automatically on your defined schedule.

Cron expressions for precise scheduling (e.g., daily at 9 AM, weekdays only, every 30 minutes). Interval for simple repeating schedules (run every X minutes, minimum 1 minute).

Schedules run on FlowDot's servers, not your computer. They execute even if you're offline or your computer is turned off.

Yes! Click the Power icon on any schedule to pause it. The configuration is preserved, and no executions occur while paused. Click again to resume.

Click the Clock icon on any schedule to see execution logs with status (success/failed/skipped), start and end times, duration, and error messages if any.

Enterprise-grade encryption for data in transit and at rest. All credentials are encrypted in the database. Workflows don't contain actual key values, and keys are never visible in execution logs.

Private workflows: Only you and invited team members. Public workflows: Anyone can view and copy them from the Public Workflow Browser.

Yes, full GDPR compliance with data portability and deletion rights. You can export your data and delete your account at any time.

Encrypted storage with secure key management practices. Keys are user-specific, never shared, and substituted at runtime so they're never exposed in workflow definitions.

Yes. Memories are completely private to your account. Personal documents in the Knowledge Base are only visible to you. Team documents are visible only to team members.

Check the gas meter icon in the header for quick status (green/yellow/red). View detailed usage on your Profile page showing executions remaining, tokens remaining, and plan tier.

Each time you run a workflow counts as one execution, regardless of workflow complexity or number of nodes. Free tier: 5 executions reset daily. Creator: 1,000/month. Pro: 5,000/month. Business: 25,000/month.

Workflows pause until your limit resets (daily for Free executions/tokens, hourly for Creator tokens, monthly for paid executions and Pro/Business tokens) or you upgrade. There are no overage charges. You can still use your own API keys for LLM usage.

Executions: How many times you run workflows. Tokens: AI processing units consumed by LLM nodes using FlowDot's internal models. They're separate limits.

Navigate to your Profile, find the Subscription Plan section, click Upgrade Plan, compare plans, select your preferred tier, and complete checkout. Monthly and yearly billing options available.

It depends on the resource and plan. Executions: Free resets daily, paid plans reset monthly. Tokens: Free resets daily, Creator resets hourly (250K/hour), Pro and Business reset monthly. Check https://flowdot.ai/subscriptions for details.

FlowDot Credits let you use 30+ AI providers through a single billing system. We handle provider relationships, API management, and billing complexity. Credits are consumed based on actual token usage across all providers. This eliminates needing multiple subscriptions and provides unified usage tracking.

BYOK (Bring Your Own Keys) is always most cost-effective since you pay providers directly with no markup. FlowDot Credits add a small margin for convenience but offer benefits like unified billing, no minimum commits, and instant access to all providers. Use BYOK for high-volume production, Credits for exploration and development.

Yes! You can use BYOK for your primary models and Credits for providers you use occasionally. The system automatically routes to the appropriate billing method. This hybrid approach optimizes costs while maintaining flexibility to experiment with different models.

Click the hamburger menu, select Teams, click Create Team, enter a team name and optional description, then click Create. You become the team Owner with full control.

Viewer: View and run workflows. Editor: Plus create/edit workflows and upload documents. Admin: Plus invite/remove members and change roles. Owner: Full control including team deletion.

Open your team's detail page, click Invite Member, enter their email address, select their role (Viewer, Editor, or Admin), and send the invitation. Share the generated link with them.

Yes! You can be a member of as many teams as needed, with different roles in each team.

Create the workflow within the team (select team when creating), or move an existing personal workflow to the team. Team workflows are visible to all team members based on their role permissions.

No. Personal workflows are private to you. Only workflows created within the team are shared with team members.

Check: execution logs for errors, API keys are configured correctly, you're within usage limits, all required inputs are provided, and nodes are properly connected.

Common causes: corrupted file, image-only PDF (no extractable text), unsupported encoding, password-protected PDF, or file too large. Try re-uploading or check the file format. Click retry icon to reprocess.

Check that: the toolkit is installed, it's active (not paused), all required credentials are mapped, and tool descriptions clearly explain what they do.

Ensure Ollama is running (ollama serve), the URL is correct (default: http://localhost:11434), your browser allows localhost connections, and you have models downloaded (ollama list).

Documentation at https://flowdot.ai/learning-center, community forums, and support tickets. Response time varies by plan: Free (community support), Paid (24-48 hours), Business (priority support).

In API Settings, after entering a key, click "Test API Key". You'll see a success or error message indicating whether the key is valid.

Check that: the integration is set to Active, the webhook URL is correct in the external service, the event type matches your mapping, and credentials are still valid.

Memories are personal notes that help FlowDot's AI understand your preferences and context. They personalize your interactions with the AI agent. Categories include General, Preferences, Facts, and Behaviors.

Visit another user's public profile and click Follow. Their public content appears in your network. You can manage followers and following from your profile settings.

MCP (Model Context Protocol) tokens are for integrations like ChatGPT MCP connections. Create them in Security Settings, select permissions, and save the generated token securely (it's only shown once).

LLM API Keys: For AI model providers (OpenAI, Claude, Gemini, etc.). User API Keys: For any other service (weather APIs, databases, custom integrations). Use naming like MY_WEATHER_API.

Reference keys in text fields using double curly braces: https://api.example.com?api_key={{MY_API_KEY}}. The key is automatically substituted at runtime.

In the Editor, click the Run button (play icon) in the toolbar. In the Workflow Runner, fill in any required inputs and click Execute. You can also run workflows through the AI Agent by describing what you want to do.

Results appear in output nodes after execution completes. You can also view the Debug Panel to see data flowing through each node, and check the Execution History for past runs.

Yes! During execution, nodes show visual indicators: blue pulse means processing, green means complete, red means error. The Debug Panel shows real-time data flowing through each node.

Failed nodes show red indicators. Check the Debug Panel or execution logs for error messages. Common issues include missing API keys, invalid inputs, or connection timeouts. Fix the issue and re-run.

Yes! Access Execution History from the workflow to see past runs with timestamps, status (success/failed), duration, inputs used, and outputs generated. You can reload results from history to view or share them.

Nodes are visual blocks representing actions (AI prompts, integrations, data processing). Each node has input sockets (left side) to receive data and output sockets (right side) to send data. Connect them to create workflows that process data step-by-step.

Click on an output socket (right side of a node) and drag to an input socket (left side of another node) to create a connection. Data flows from outputs to inputs through these connections.

Input nodes (text, file, image, slider), Output nodes (text display, debug), LLM nodes (GPT, Claude, Gemini, local models), Integration nodes (Discord, Slack, GitHub), Logic nodes, Data transformation nodes, RAG/Knowledge Base nodes, and Custom nodes.

Click on any node to open the Properties Panel on the right side of the editor. Here you can configure all settings specific to that node type, such as prompts for LLM nodes, API settings, or input defaults.

Pan the canvas by dragging with the middle mouse button or holding spacebar while dragging. Zoom using the mouse wheel or zoom controls. Use the minimap (bottom-right) for quick navigation on large workflows.

Yes! Clone any public workflow from the Public Workflow Browser by clicking "Copy to My Workflows", or duplicate your own workflows as starting points.

Open Workflow Settings (gear icon), toggle "Make workflow public" to add it to the Public Workflow Browser where anyone can discover and copy it. For team-only sharing, keep it private and share within your team.

Select the node and press Delete/Backspace, or right-click and choose Delete. For connections, click on the connection line and press Delete, or drag from the input socket to disconnect.

Socket colors indicate data types - they help you understand what kind of data flows through connections. Matching colors generally mean compatible types, though some connections work across different types.

FlowDot uses three model tiers: Simple (fast, cheap - GPT-4o-mini, Claude Haiku), Capable (balanced - GPT-4o, Claude Sonnet), and Complex (maximum intelligence - o1, Claude Opus). Nodes can specify tier requirements, and the system auto-selects appropriate models. This optimizes cost and performance based on task complexity.

After workflow execution, click "Continue Conversation" to maintain context and chat with the AI about results. The system preserves the execution state, allowing follow-up questions, refinements, or additional processing without re-running the entire workflow. This is perfect for iterative tasks or exploring results.

The editor provides real-time validation, checking for disconnected nodes, type mismatches, missing required inputs, and circular dependencies. Red highlights indicate errors, yellow shows warnings. Hover over issues for detailed explanations. This prevents runtime failures and ensures workflows execute successfully.

Installation

## Installation

### npm (recommended)
```bash
npm install -g @flowdot/cli
```

### yarn
```bash
yarn global add @flowdot/cli
```

### From source
```bash
git clone https://github.com/flowdot/cli
cd cli
npm install
npm run build
npm link
```

Quick Start

1. Authenticate

flowdot auth login

This opens your browser for secure authentication.

2. List your workflows

flowdot workflows list

3. Run a workflow

flowdot run <workflow-hash>

4. Run with inputs

flowdot run wf_abc123 --input document_url="https://example.com/doc.pdf"

5. Interactive agent mode

flowdot agent

Source repository ↗ Package ↗

Installation

## Installation

### npm
```bash
npm install -g @flowdot.ai/mcp-server
```

### npx (no install)
```bash
npx @flowdot.ai/mcp-server
```

## Requirements

- Node.js >= 18.0.0
- FlowDot MCP Token (get from Settings → MCP Tokens)

Quick Start

1. Get an MCP Token

1. Log into FlowDot
2. Go to Settings → MCP Tokens
3. Create a new token with desired scopes
4. Copy the token (starts with fd_mcp_)

2. Configure Your AI Assistant

Claude Desktop

Add to your Claude config file:

{
  "mcpServers": {
    "flowdot": {
      "command": "npx",
      "args": ["@flowdot.ai/mcp-server"],
      "env": {
        "FLOWDOT_API_TOKEN": "fd_mcp_your_token_here"
      }
    }
  }
}

3. Use FlowDot Tools

Once configured, your AI assistant can:

• List and execute your workflows
• Chat with the FlowDot Agent
• Create and manage workflows
• Search public workflows
• And more...

Source repository ↗ Package ↗

Installation

## Installation

```bash
npm install @flowdot/mobile-sdk
# or
yarn add @flowdot/mobile-sdk
```

### iOS Setup
```bash
cd ios && pod install
```

### Android Setup
No additional setup required.

Quick Start

import { FlowDotClient } from '@flowdot/mobile-sdk';

// Initialize
const client = new FlowDotClient({
  apiUrl: 'https://flowdot.ai/api',
});

// Authenticate
await client.login(email, password);

// List workflows
const workflows = await client.getWorkflows();

// Execute workflow
const result = await client.executeWorkflow(workflowId, {
  inputs: { message: 'Hello' }
});

Installation

## Installation

```bash
pip install openai
```

Requires Python 3.8+.

Quick Start

from openai import OpenAI

client = OpenAI(
    api_key="fd_agg_xxxxxxxxxxxxx",
    base_url="https://flowdot.ai/api/v1",
)

resp = client.chat.completions.create(
    model="openai/gpt-4o-mini",  # or anthropic/claude-sonnet-4-6, redpill/google/gemini-2.5-flash-lite, etc.
    messages=[{"role": "user", "content": "Hello from FlowDot"}],
)
print(resp.choices[0].message.content)

Streaming, tool calling, embeddings, and audio all work identically to the OpenAI API.

Source repository ↗ Package ↗

Installation

## Installation

```bash
npm install openai
```

Works in Node.js 18+, Deno, Bun, and Cloudflare Workers.

Quick Start

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.FLOWDOT_AGG_KEY,
  baseURL: 'https://flowdot.ai/api/v1',
});

const resp = await client.chat.completions.create({
  model: 'anthropic/claude-sonnet-4-6',
  messages: [{ role: 'user', content: 'Hello from FlowDot' }],
});
console.log(resp.choices[0].message.content);

Streaming:

const stream = await client.chat.completions.create({
  model: 'redpill/google/gemini-2.5-flash-lite',
  messages: [{ role: 'user', content: 'Tell me a bee fact' }],
  stream: true,
});
for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? '');
}

Source repository ↗ Package ↗

Major platform update with new features and improvements.

Highlights

• Live Agent Voice Mode: Real-time voice conversations with streaming TTS/STT
• Agent Toolkits (MCP): Create and install tool collections for the AI Agent
• Multi-File Apps: Organize React apps across multiple files
• Team Knowledge Bases: Share documents with team members
• Fork Lineage Tracking: See the full history of copied workflows
• Energy Transparency: Track AI energy consumption in workflows

Introducing FlowDot Apps and enhanced Custom Nodes.

New Features

• FlowDot Apps: Build custom React frontends powered by workflows
• App Display Modes: Windowed, fullscreen, and embedded options
• Custom Node LLM Support: Call AI from custom node scripts
• Public Workflow Browser: Discover and copy community workflows

First public release of FlowDot.

Core Features

• Visual workflow editor with 60+ node types
• AI Agent with conversation interface
• MCP integration for AI assistants
• Knowledge Base with RAG support
• Team collaboration
• Scheduling and automation
• Discord, Slack, and GitHub integrations