🧵 Fabric Mini GUI

A modern desktop client for interacting with Fabric AI patterns
Built with CustomTkinter for a sleek, dark-mode friendly experience

Overview

Fabric Mini GUI provides a graphical desktop interface for the Fabric AI framework by Daniel Miessler. Instead of using the Fabric CLI directly, you get a modern dark-mode window where you can:

Browse and search AI patterns visually
Select from all your configured AI models (Claude, GPT-4, Ollama, etc.)
Stream AI responses in real-time
Manage the Fabric server with start/stop controls and live health monitoring
Navigate through your response history

The entire application is a single Python file (fabricgui.py) with only two dependencies: requests and customtkinter.

Architecture

System Architecture

High-level view of how Fabric GUI interacts with the Fabric ecosystem:

graph TB
    subgraph FabricGUI["🖥️ Fabric GUI Application"]
        GUI["FabricGUI<br/>(Main Window)"]
        CM["ConfigManager"]
        OH["OutputHistory"]
        SM["ServerManager"]
        PD["PreferencesDialog"]
        CTX["ContextMenu"]
    end

    subgraph FabricCLI["⚙️ Fabric CLI"]
        SERVE["fabric --serve"]
        PATTERN["fabric -p pattern"]
    end

    subgraph Providers["☁️ AI Providers"]
        CLAUDE["Claude"]
        GPT["GPT-4"]
        OLLAMA["Ollama"]
        OTHER["Other LLMs"]
    end

    subgraph Storage["💾 Local Storage"]
        CFG["~/.fabric_gui/config.json"]
        HIST["~/.fabric_gui/history.json"]
        LOG["~/.fabric_gui/fabric_gui.log"]
        FENV["~/.config/fabric/.env"]
    end

    GUI -->|"Manages"| SM
    GUI -->|"Reads/Writes"| CM
    GUI -->|"Navigates"| OH
    GUI -->|"Opens"| PD
    GUI -->|"Attaches to text widgets"| CTX

    SM -->|"Starts/Stops"| SERVE
    SM -->|"GET /patterns/names"| SERVE
    SM -->|"GET /models/names"| SERVE
    SM -->|"Health checks"| SERVE

    GUI -->|"Spawns subprocess"| PATTERN
    PATTERN -->|"stdin: user text"| PATTERN
    PATTERN -->|"stdout: AI response"| GUI

    SERVE -->|"Routes to"| CLAUDE
    SERVE -->|"Routes to"| GPT
    SERVE -->|"Routes to"| OLLAMA
    SERVE -->|"Routes to"| OTHER

    CM -->|"Persists"| CFG
    OH -->|"Persists"| HIST
    GUI -->|"Logs to"| LOG
    SM -->|"Reads default model"| FENV

    style FabricGUI fill:#1a1a2e,stroke:#16213e,color:#e94560
    style FabricCLI fill:#0f3460,stroke:#16213e,color:#e94560
    style Providers fill:#533483,stroke:#16213e,color:#e94560
    style Storage fill:#2b2b2b,stroke:#444,color:#ccc

Class Diagram

Detailed view of all classes, their fields, and relationships:

classDiagram
    class ConfigManager {
        <<static>>
        +Path CONFIG_FILE
        +dict DEFAULT_CONFIG
        +load() Dict~str, Any~
        +save(config: Dict) void
    }

    class OutputHistory {
        +Path HISTORY_FILE
        -List~Dict~ history
        -int max_size
        -int current_index
        +__init__(max_size: int)
        +add(pattern, input_text, output_text)
        +update_current_output(output_text)
        +previous() Optional~Dict~
        +next() Optional~Dict~
        +has_previous() bool
        +has_next() bool
        +load() void
        +save() void
    }

    class ServerManager {
        -str fabric_command
        -str base_url
        -str port_flag
        -Popen process
        -bool is_online
        -bool _monitoring
        -Thread _health_thread
        +__init__(fabric_command, base_url, port_flag)
        +check_health() bool
        +start_health_monitoring(interval, callback)
        +stop_health_monitoring()
        +start_server() bool
        +stop_server(timeout) bool
        +get_patterns() Optional~List~
        +get_models() Dict~str, List~
        +get_default_model() Optional~str~
        +is_running() bool
        -_normalize_base_url(url) str
        -_port_from_base_url(base_url) str
        -_start_server_output_capture()
    }

    class ContextMenu {
        -Widget widget
        -Menu menu
        +__init__(widget: Widget)
        -_gen(ev: str) void
        -_select_all() void
        -_show(event) void
    }

    class PreferencesDialog {
        -CTk parent
        -Dict config
        -Optional~Dict~ result
        -StringVar base_url_var
        -BooleanVar auto_start_var
        -BooleanVar stop_on_exit_var
        -StringVar health_interval_var
        -StringVar fabric_cmd_var
        -StringVar timeout_var
        +__init__(parent, config)
        -_build_ui()
        -_build_server_tab()
        -_build_advanced_tab()
        -_validate_and_collect() Optional~Dict~
        -_on_save()
        -_on_cancel()
    }

    class FabricGUI {
        -Dict app_config
        -OutputHistory history
        -ServerManager server_manager
        -bool cancel_request
        -Thread current_request_thread
        -Popen current_process
        -StringVar base_url_var
        -StringVar pattern_var
        -StringVar model_var
        -StringVar status_var
        -StringVar command_var
        -List~str~ all_patterns
        +__init__()
        -_build_menu()
        -_build_server_frame()
        -_build_pattern_frame()
        -_build_info_frame()
        -_build_io_frame()
        -_setup_shortcuts()
        +on_send(event)
        +on_cancel()
        +on_start_server()
        +on_stop_server()
        +on_test_server()
        +load_patterns()
        +load_models()
        +save_output()
        +copy_output()
        +clear_output()
        +import_file()
        +show_preferences()
        +show_help()
        +on_closing()
        -_process_request(input_text)
        -_filter_patterns(*args)
        -_update_led_status(is_online)
        -_set_ui_processing(processing)
        -_animate_progress()
    }

    FabricGUI --> ConfigManager : loads/saves config
    FabricGUI --> OutputHistory : manages history
    FabricGUI --> ServerManager : controls server
    FabricGUI --> ContextMenu : attaches to text widgets
    FabricGUI --> PreferencesDialog : opens dialog
    FabricGUI --|> CTk : inherits
    PreferencesDialog --|> CTkToplevel : inherits

Request Processing Flow

Step-by-step view of what happens when the user clicks Send:

flowchart TD
    A["👤 User clicks Send<br/>(or Ctrl+Enter)"] --> B{"Pattern<br/>selected?"}
    B -->|No| C["⚠️ Show warning:<br/>Please select a pattern"]
    B -->|Yes| D{"Server<br/>online?"}

    D -->|No| E["🔌 Prompt: Start server?"]
    E -->|No| F["❌ Cancel"]
    E -->|Yes| G["🚀 Start server<br/>Wait for health check"]
    G --> H

    D -->|Yes| H["💾 Save config<br/>(pattern, model, URL)"]
    H --> I["🔧 Build command:<br/>fabric -p pattern [-m model]"]
    I --> J["🔒 Disable Send button<br/>Enable Cancel button<br/>Start progress animation"]
    J --> K["📤 Spawn subprocess<br/>(CREATE_NO_WINDOW on Windows)"]
    K --> L["📝 Write input text to stdin<br/>(in background thread)"]
    L --> M["📖 Add history entry"]

    M --> N{"Read stdout<br/>chunk"}
    N -->|Chunk received| O{"Cancel<br/>requested?"}
    O -->|Yes| P["🛑 Terminate process"]
    O -->|No| Q{"Filter<br/>line?"}
    Q -->|Yes: Ollama noise| N
    Q -->|No| R["📺 Append to output<br/>(via self.after)"]
    R --> N

    N -->|EOF + process done| S["✅ Update status:<br/>Completed"]
    S --> T["💾 Save history"]
    T --> U["🔓 Re-enable Send<br/>Disable Cancel<br/>Stop animation"]

    P --> V["📝 Set status: Cancelled"]
    V --> U

    style A fill:#1f6aa5,color:#fff
    style C fill:#cc6600,color:#fff
    style F fill:#cc3333,color:#fff
    style P fill:#cc3333,color:#fff
    style S fill:#2d8f2d,color:#fff

Server Lifecycle

How ServerManager controls the Fabric server process:

stateDiagram-v2
    [*] --> Offline : App starts

    Offline --> Starting : on_start_server()
    Starting --> Online : Health check passes
    Starting --> Offline : Process exited immediately

    Online --> Online : Health check (every 5s) ✓
    Online --> Offline : Health check fails
    Online --> Stopping : on_stop_server()

    Stopping --> Offline : Process terminated
    Stopping --> Offline : Force kill (after timeout)

    Offline --> Online : Auto-start on launch
    Offline --> Prompted : Send request while offline
    Prompted --> Starting : User says "Yes"
    Prompted --> Offline : User says "No"

    state Online {
        [*] --> Healthy
        Healthy --> Healthy : GET /readyz → 200
        Healthy --> Degraded : Request fails
        Degraded --> Healthy : Next check passes
    }

    note right of Starting
        Command: fabric --serve --address :PORT
        Creates new process group (Windows)
        Captures stdout in background thread
        Waits 2s for process stability
    end note

    note right of Online
        LED indicator: 🟢
        Patterns auto-loaded
        Models fetched
    end note

    note right of Offline
        LED indicator: 🔴
        Server buttons update
    end note

Application Startup Sequence

What happens when the application launches:

sequenceDiagram
    participant User
    participant GUI as FabricGUI.__init__
    participant CM as ConfigManager
    participant SM as ServerManager
    participant FS as File System

    User->>GUI: python fabricgui.py

    GUI->>CM: load()
    CM->>FS: Read ~/.fabric_gui/config.json
    FS-->>CM: Config dict (or defaults)
    CM-->>GUI: app_config

    GUI->>GUI: OutputHistory()
    GUI->>FS: Read ~/.fabric_gui/history.json

    GUI->>SM: ServerManager(fabric_cmd, base_url, port_flag)

    GUI->>GUI: Build UI components
    Note over GUI: _build_menu()<br/>_build_server_frame()<br/>_build_pattern_frame()<br/>_build_info_frame()<br/>_build_io_frame()<br/>_setup_shortcuts()

    GUI->>SM: start_health_monitoring(interval=5s)
    Note over SM: Background thread begins<br/>GET /readyz every 5s

    alt auto_start_server = true
        GUI->>SM: start_server() (after 600ms)
        SM->>SM: fabric --serve --address :PORT
    end

    GUI->>SM: load_patterns() (after 800ms)
    SM->>SM: GET /patterns/names
    SM-->>GUI: Pattern list → dropdown

    GUI->>SM: load_models() (after 1200ms)
    SM->>SM: fabric --listmodels
    SM-->>GUI: Models by provider → dropdown
    SM->>FS: Read ~/.config/fabric/.env
    FS-->>SM: DEFAULT_MODEL
    SM-->>GUI: Default model label

    GUI-->>User: Window ready

GUI Layout Structure

Visual map of how the UI panels are organized:

block-beta
    columns 1

    block:top["Server Frame"]
        columns 6
        LED["🔴 LED"] URL["Base URL Label"] Test["Test"] Start["Start"] Stop["Stop"] space
    end

    block:pattern["Pattern Selection Frame"]
        columns 4
        Search["🔍 Search Input"] space:3
        PatternLabel["Pattern:"] PatternCombo["Pattern Dropdown ▼"] Refresh["Refresh Patterns"] space
        ModelLabel["Model:"] ModelCombo["Model Dropdown ▼"] DefaultModel["Default: model-name"] space
    end

    block:info["Info / Actions Frame"]
        columns 3
        Status["Status: Ready"] CommandPreview["Command: fabric -p pattern"] Actions["Cancel | Send"]
    end

    block:io["Input / Output Frame"]
        columns 2
        block:input["Input Panel"]
            InputToolbar["Import | Paste | Clear"]
            InputText["📝 Input Textbox"]
        end
        block:output["Output Panel"]
            OutputToolbar["Copy | Save | Clear | ◀ ▶"]
            OutputText["📄 Output Textbox"]
        end
    end

Features

✨ Core Functionality

Pattern selection and execution via Fabric CLI subprocess
Real-time streaming responses — output appears chunk-by-chunk as the AI generates it
Model selection — browse all configured AI models grouped by provider
Command preview — see the exact fabric command before executing

🎨 Modern User Interface

CustomTkinter integration for a modern, dark-mode friendly aesthetic
Clean, intuitive layout with organized frames
Animated progress indicator with pulsing gold dots during processing
Output toolbar (Copy, Save, Clear)
History navigation (◀/▶ buttons)
Status bar with visual feedback
Context Menus — right-click support for Cut/Copy/Paste/Select All
Interactive Pattern Search — real-time filtering with match count display

🖥️ Server Management

Visual LED status indicator (🔴 offline / 🟢 online)
Start/Stop server controls directly from the GUI
Automatic health monitoring (configurable interval, default 5s)
Auto-Load Patterns — patterns and models load automatically when server comes online
Pre-request server validation with auto-start prompt
Graceful shutdown handling with force-kill fallback

⚙️ Configuration

Persistent settings via JSON (auto-saved)
Tabbed Preferences dialog (Server + Advanced)
Configurable server URL, health check interval, and request timeout
Window geometry persistence across sessions

📝 Output Management

Copy output to clipboard
Save output to file (TXT/MD)
Clear output display
Navigate through response history (up to 50 entries, persisted to disk)

📁 File Import

Import .txt or .md files directly into the input box via the Import button

Installation

Requirements

Python 3.11+
Fabric CLI installed and on PATH
requests library
customtkinter library

Quick Start (Windows)

Clone or download this repository
Double-click start.bat
- First run: creates a virtual environment, installs dependencies, and launches the app
- Subsequent runs: activates the venv and launches immediately

Manual Setup

# Clone the repository
git clone https://github.com/Digitalgods2/FabricGui.git
cd FabricGui

# Install dependencies
pip install -r requirements.txt

# Run the application
python fabricgui.py

Build Standalone Executable

pip install pyinstaller
pyinstaller fabricgui.spec

The standalone FabricGUI.exe will be created in the dist/ folder (no console window).

Usage

First Time Setup

Configure Server
- The default server URL is http://localhost:8083
- Open Edit → Preferences to change the URL or other settings
- Click Test to verify connectivity
Start the Server
- Click Start to launch Fabric's built-in server
- Wait for the LED indicator to turn 🟢 green
- Patterns and models load automatically
Select Your Model
- The Model dropdown shows all available AI models grouped by provider
- Click the "Default: ..." label to reset to your configured default model

Sending Requests

Select a pattern from the dropdown (use the search box to filter)
Enter your input text (or click Import to load a file)
Review the command preview
Click Send or press Ctrl+Enter
Watch the response stream in the output panel

Managing Output

Action	Button	Shortcut
Copy to clipboard	Copy	`Ctrl+C`
Save to file	Save	`Ctrl+S`
Clear display	Clear	—
Previous history	◀	`Alt+Left`
Next history	▶	`Alt+Right`

Managing the Server

Status Indicators:

LED	Meaning
🔴 Red	Server offline
🟢 Green	Server online

Auto-Start: Enable in Preferences to start the server automatically on app launch.

Pre-Request Validation: If the server is offline when you send a request, the app prompts: "Would you like to start it now?"

Cancelling Requests

Click the Cancel button during processing. The subprocess will be terminated gracefully (or force-killed after a timeout).

Configuration

Settings are stored at ~/.fabric_gui/config.json and auto-saved:

{
  "base_url": "http://localhost:8083",
  "last_pattern": "",
  "last_model": "",
  "window_geometry": "1200x800",
  "auto_start_server": false,
  "stop_server_on_exit": true,
  "server_health_check_interval": 5,
  "request_timeout": 300,
  "fabric_command": "fabric",
  "port_flag": "--address"
}

Setting	Default	Description
`base_url`	`http://localhost:8083`	Fabric server address
`auto_start_server`	`false`	Launch server on app start
`stop_server_on_exit`	`true`	Prompt to stop server on close
`server_health_check_interval`	`5`	Health check frequency (seconds)
`request_timeout`	`300`	Max processing time (seconds)
`fabric_command`	`fabric`	Path to Fabric executable
`port_flag`	`--address`	Server bind flag (auto-corrected from `--port`)

Keyboard Shortcuts

Shortcut	Action
`Ctrl+Enter`	Send request
`Ctrl+S`	Save output to file
`Ctrl+C`	Copy output to clipboard
`Ctrl+,`	Open Preferences
`Alt+Left`	Previous history entry
`Alt+Right`	Next history entry
Right-click	Context menu (Cut/Copy/Paste/Select All)

Menu Reference

Menu	Item	Shortcut	Action
File	Save Output...	`Ctrl+S`	Save output to file
	Exit	—	Close application
Edit	Preferences...	`Ctrl+,`	Open settings dialog
	Copy Output	`Ctrl+C`	Copy to clipboard
	Clear Output	—	Clear output display
	Paste Input	—	Paste from clipboard
	Clear Input	—	Clear input text
History	Previous	`Alt+Left`	Previous response
	Next	`Alt+Right`	Next response
Help	User Guide	—	Comprehensive help window
	View Logs	—	Open log file
	About	—	Application info

Project Structure

FabricGui/
├── fabricgui.py          # Entire application (1,930 lines)
├── fabricgui.spec        # PyInstaller build configuration
├── requirements.txt      # Python dependencies
├── start.bat             # Windows quick-start launcher
├── README.md             # This file
├── .gitignore
└── build/                # PyInstaller build artifacts

Runtime files (created automatically):
~/.fabric_gui/
├── config.json           # Persistent configuration
├── history.json          # Response history (up to 50 entries)
└── fabric_gui.log        # Rotating log files (5 MB × 3 backups)

Key Classes

Class	Lines	Responsibility
`ConfigManager`	248–314	Loads/saves JSON config with defaults and migration
`OutputHistory`	321–387	Manages response history with persistence
`ServerManager`	394–660	Controls Fabric server process, health checks, pattern/model fetching
`ContextMenu`	667–714	Right-click Cut/Copy/Paste/Select All for any text widget
`PreferencesDialog`	721–898	Tabbed settings dialog with validation
`FabricGUI`	905–1924	Main window — all UI building, event handling, and request processing

Logging

Application logs are saved to ~/.fabric_gui/fabric_gui.log

Max file size: 5 MB per file
Backup files: 3 rotated backups
Total max: ~20 MB
View logs: Help → View Logs

Troubleshooting

"Failed to reach server"

Verify the Fabric server is running (LED should be 🟢)
Check the base URL in Preferences
Click Test to diagnose connectivity

"Error loading patterns"

Ensure the server is accessible and responding
Check API key if required
Review logs (Help → View Logs) for detailed error messages

Request Timeout

Increase timeout in Edit → Preferences → Advanced
Check network connection
Verify the server is responding (click Test)

"Ollama Get ... connection refused"

This is normal if you are not running a local Ollama instance
The Fabric CLI automatically checks for local models on startup
The GUI filters this message to keep output clean
It does not affect cloud models (Claude, GPT-4, etc.)

Server Won't Start

Ensure fabric is installed and on your system PATH
Try running fabric --serve in a terminal to see errors directly
Check that the port (default 8083) is not already in use

Version History

Version 3.2 (Latest) 🚀

✅ Interactive Pattern Search: Real-time filtering with auto-selection of first match
✅ Improved Readability: Enhanced dropdown styling with better contrast, taller dropdowns (25 items)
✅ UI Polish: Consistent font sizing, status bar shows match count while searching

Version 3.1

✅ Reliable AI Processing: Switched to direct subprocess execution for guaranteed correct output
✅ Enhanced UI: Vertical scrollbar on pattern dropdown, expanded list (40 items), increased fonts
✅ Smarter Server Management: Improved stop logic, better error handling
✅ Bug Fixes: Output buffering/hanging, encoding issues with emojis, filtered startup errors

Version 3.0 ⭐ Major Update

✅ CustomTkinter Migration: Complete UI overhaul with modern look and dark mode
✅ Context Menus: Right-click support for text widgets
✅ Bug Fixes: Infinite pattern loading loop, endpoint fix, TypeError, startup crashes

Version 2.1

✅ Server management with Start/Stop controls and LED indicator
✅ Automatic health monitoring and pre-request validation
✅ Auto-start server option and graceful shutdown
✅ Cross-platform process management

Version 2.0

✅ Complete refactoring with improved architecture
✅ Configuration persistence, output history, request cancellation
✅ Progress indicators, menu system, keyboard shortcuts
✅ Comprehensive logging

Version 1.0

Basic Fabric pattern execution with input/output interface

Contributing

Suggestions and improvements are welcome! Please ensure:

Code follows existing style
All features are tested
Documentation is updated

Credits

Fabric GUI designed and developed by DigitalGods.ai

Built for the Fabric AI framework by Daniel Miessler.

Happy pattern processing! 🚀

Name		Name	Last commit message	Last commit date
Latest commit History 24 Commits
.vscode		.vscode
build/fabricgui		build/fabricgui
.gitignore		.gitignore
README.md		README.md
fabricgui.py		fabricgui.py
fabricgui.spec		fabricgui.spec
requirements.txt		requirements.txt
start.bat		start.bat

Folders and files

Latest commit

History

Repository files navigation

🧵 Fabric Mini GUI

Table of Contents

Overview

Architecture

System Architecture

Class Diagram

Request Processing Flow

Server Lifecycle

Application Startup Sequence

GUI Layout Structure

Features

✨ Core Functionality

🎨 Modern User Interface

🖥️ Server Management

⚙️ Configuration

📝 Output Management

📁 File Import

Installation

Requirements

Quick Start (Windows)

Manual Setup

Build Standalone Executable

Usage

First Time Setup

Sending Requests

Managing Output

Managing the Server

Cancelling Requests

Configuration

Keyboard Shortcuts

Menu Reference

Project Structure

Key Classes

Logging

Troubleshooting

"Failed to reach server"

"Error loading patterns"

Request Timeout

"Ollama Get ... connection refused"

Server Won't Start

Version History

Version 3.2 (Latest) 🚀

Version 3.1

Version 3.0 ⭐ Major Update

Version 2.1

Version 2.0

Version 1.0

Contributing

Credits

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Uh oh!

Contributors

Uh oh!

Languages

Packages