FlowMason VSCode Extension

The official VSCode extension for FlowMason - Universal AI Workflow Infrastructure.

Build, debug, and deploy AI pipelines directly in VSCode with a Salesforce DX-style workflow.

What's New in v0.9.3

• Fixed Lab Component Registration - Critical bug fix
  • Studio now properly auto-discovers built-in lab components
  • Fixed issue where old cached packages would override updated components
  • json_transform now correctly parses JSON strings with parse_json_strings=True

What's New in v0.9.2

• Fixed Pipeline Execution from VSCode - Critical bug fix
  • Pipelines with template variables (e.g., {{upstream.stage.field}}) now work correctly
  • Fixed field mapping in Studio API that was causing template resolution errors
  • CLI and VSCode/Studio now use consistent execution paths

What's New in v0.9.1

• Fixed Studio Startup - Now uses fm studio start CLI command
  • Works correctly with pip-installed FlowMason
  • No longer requires source code or npm
  • Studio management is simpler and more reliable

What's New in v0.9.0

• Control Flow Components Now Fully Working

  • ForEach Loops - Iterate over collections with {{context.item_variable}} and {{context.index_variable}}
  • TryCatch Error Handling - Try/catch blocks with proper nested result propagation
  • Conditional Branching - Route execution based on conditions
  • Router (Switch/Case) - Multi-branch routing based on values
  • All control flow components properly integrate with the DAG executor

• Loop Variable Injection - Access current item and index in foreach loops:

  {
    "id": "transform-item",
    "config": {
      "data": {
        "item": "{{context.current_item}}",
        "index": "{{context.item_index}}"
      }
    },
    "depends_on": ["foreach-stage"]
  }
  

• Nested Result Propagation - Access results from try/catch nested stages:

  {
    "id": "process-result",
    "config": {
      "data": {
        "api_result": "{{upstream.fetch-api}}",
        "status": "{{upstream.trycatch-stage.status}}"
      }
    },
    "depends_on": ["trycatch-stage"]
  }
  

What's New in v0.8.0

• Visual-First Configuration Editor (Salesforce Flow-style)
  • No more typing {{...}} syntax - everything through dropdowns
  • Source Picker: Dropdown to select data sources (Pipeline Input, upstream stages)
  • Visual Field Mapper: Table UI to map input fields to output fields for json_transform
  • Prompt Editor: Click variable pills to insert references into prompts
  • Generated JSON View: Read-only, collapsible section showing the actual config
  • Configuration is 100% visual - JSON is just output for CI/CD

What's New in v0.7.1

• Smart Data Source Awareness - Configuration editor now understands data flow
  • Shows available Pipeline Input fields from input_schema
  • Shows upstream stages with their output fields
  • Click field chips to insert correct variable syntax automatically
  • Visual field picker for {{input.field}} and {{stage.result.field}} references

What's New in v0.7.0

• Enhanced Configuration Editor - Beautiful full-panel editor for stage configuration
  • Opens on the right side with more space
  • Shows all available fields for each component with descriptions
  • Dropdowns for enum fields, checkboxes for booleans
  • Number inputs with min/max/step validation
  • JSON editors for complex objects
  • Built-in schemas for all core components (generator, json_transform, http_request, etc.)
  • Tips section showing variable syntax
  • Category badges and icons for visual clarity
• Fix - Circular dependency prevention - cannot create cycles in DAG
• Fix - Stage Config panel context now properly set

What's New in v0.6.11

• Fix - Stage Config panel now shows when DAG canvas is open
• Fix - Circular dependency prevention - cannot create connections that would form cycles
• Improved - Warning message when attempting to create invalid circular connection

What's New in v0.6.10

• Fix - "Edit Configuration" now works from DAG canvas right-click menu (opens Stage Config panel)
• Fix - Keyboard shortcuts (A, F, Delete, etc.) now work in DAG canvas
• Improved - Stage Config panel automatically focuses when editing configuration

What's New in v0.6.9

• Fix - Fixed node dragging not working (positions now update correctly during drag)
• Fix - Fixed connection ports not responding to clicks
• Improved - Input ports now visible when connecting for easier drop targets

What's New in v0.6.8

• Enhanced DAG Canvas - React Flow-style pipeline editor
  • Right-click context menus on nodes, edges, and canvas
  • Drag nodes to reposition (positions saved to pipeline)
  • Visual connection ports (drag from output to input)
  • Multi-select with Shift+Click or drag selection box
  • Keyboard shortcuts: Delete, Ctrl+D (duplicate), A (add), F (fit), Ctrl+A (select all)
  • Mouse wheel zoom (zoom towards cursor)
  • Minimap overview
  • Double-click node to edit config

What's New in v0.6.7

• Development Webhook Server - Built-in webhook server for testing output_config
  • Start with: FlowMason: Start Webhook Server (port 9999)
  • View received payloads in Output panel (FlowMason Webhook)
  • Configurable port via flowmason.webhookPort setting
• Fix - Add Stage command now works in DAG canvas view

What's New in v0.6.3

• Fix - Fixed pipeline template to use correct jmespath_expression field
• Fix - Fixed upstream reference format (upstream.stage.result.field)
• Fix - Added auto_discover() method to ComponentRegistry for CLI support

What's New in v0.6.2

• Fix - Fixed template type coercion error in generated pipeline (max_tokens)

What's New in v0.6.1

• Improved New Project Template - Creates Input → Generate → Output pipeline demonstrating full workflow
• Output Routing Example - New projects include output_config with webhook destination template
• Updated Dependencies - Requires FlowMason 0.5.0+

What's New in v0.6.0

• Input/Output Architecture - New output_config support in pipelines for automatic result routing
• Output Destinations - Configure webhook, email, database, and message queue destinations
• Schema Updates - Pipeline schema now supports output routing configuration
• Named Pipeline Invocation - Run pipelines by name via the API (POST /run)

What's New in v0.5.0

• New Project Command - Create complete FlowMason projects with flowmason.json, example components, and pipelines
• Updated Project Format - Uses modern flowmason.json manifest and .pipeline.json pipeline files
• Studio Management - Start, stop, and restart FlowMason Studio directly from VSCode
• Enhanced Debugging - Full debugging support with breakpoints (F9), step-through (F10), and run (F5)
• Improved Code Generation - Generated components use current API (async def execute)

Quick Start

1. Install FlowMason CLI: pip install flowmason
2. Create a Project: Cmd+Shift+P → "FlowMason: New Project"
3. Start Studio: Cmd+Shift+P → "FlowMason: Start Studio"
4. Run Pipeline: Press F5 with a .pipeline.json file open

Features

Project Management

Create and manage FlowMason projects with proper structure:

• New Project (FlowMason: New Project) - Interactive wizard creates:
  • flowmason.json - Project manifest
  • components/ - Example node and operator
  • pipelines/ - Example pipeline
  • Proper .gitignore and README.md

Component Development

Create nodes and operators with interactive wizards:

• New Node (FlowMason: New Node) - AI-powered components with LLM calls
• New Operator (FlowMason: New Operator) - Deterministic data transformations

Studio Integration

Control FlowMason Studio backend directly from VSCode:

• Start Studio - Launch the backend server (port 8999)
• Stop Studio - Gracefully stop the server
• Restart Studio - Restart with fresh state
• Status Bar - Shows connection status (Connected/Disconnected)

Development Webhook Server

Built-in webhook server for testing pipeline output routing:

• Start Webhook Server - Launch local webhook receiver (default port 9999)
• Stop Webhook Server - Stop the webhook server
• Copy Webhook URL - Copy http://localhost:9999/webhook to clipboard
• Output Panel - View received webhook payloads in real-time

Configure your pipeline's output_config to use http://localhost:9999/webhook during development. All received payloads are displayed in the "FlowMason Webhook" output panel with full request details.

Debugging

Full debugging support for pipelines:

IntelliSense & Auto-completion

Intelligent code completion for:

• @node, @operator, and @control_flow decorators
• Decorator parameters (name, description, category, etc.)
• Input/Output field definitions
• FlowMason imports

Diagnostics & Linting

Real-time validation for:

• Missing decorator parameters
• Non-kebab-case naming
• Missing Input/Output classes
• Hardcoded secrets detection
• Execute method signatures

Code Snippets

Sidebar Views

The FlowMason activity bar provides:

1. Components - Browse registered components by category
2. Pipelines - View and manage pipelines
3. Runs - Execution history

CodeLens

Clickable actions above components:

• Run - Execute the component
• Preview - View component schema
• Open in Studio - View in browser

Commands

Configuration

Requirements

• VSCode 1.85.0 or higher
• Python 3.11 or higher
• FlowMason CLI (pip install flowmason)

Installation

From VSIX

code --install-extension flowmason-0.9.3.vsix

From Source

cd vscode-extension
npm install
npm run compile
# Press F5 to launch extension development host

Project Structure

A FlowMason project created by this extension:

my-project/
├── flowmason.json           # Project manifest
├── components/
│   ├── __init__.py
│   ├── example_node.py      # AI node example
│   └── example_operator.py  # Operator example
├── pipelines/
│   └── main.pipeline.json   # Pipeline definition
├── requirements.txt
├── .gitignore
└── README.md

Links

• Documentation: https://flowmason.com/docs
• PyPI: https://pypi.org/project/flowmason/
• Support: support@flowmason.com

License

Proprietary - FlowMason