Plugins

Plugins

Plugins extend Open Ticket AI with additional capabilities. Each plugin provides services and pipes that you can use in your configuration to connect to different ticket systems, run ML models, or add custom processing logic.

What is a Plugin?

A plugin is a Python package that adds new functionality to Open Ticket AI:

• Services: Connect to ticket systems (OTOBO, Znuny), ML models, or external APIs
• Pipes: Fetch tickets, classify content, update fields, add notes, or run custom logic
• Ready to Use: Just install, configure, and reference in your YAML

How to Install Plugins

Install plugins using the uv package manager:

# Install a single plugin
uv add otai-otobo-znuny

# Install multiple plugins at once
uv add otai-otobo-znuny otai-hf-local

That’s it! Once installed, the plugin’s services and pipes are automatically available in your configuration.

How to Use Plugins in Configuration

Plugins provide services and pipes that you reference in your YAML configuration using the format plugin-name:ComponentName.

Example: Complete Workflow with Plugins

# Define services from plugins
services:
  ticket_system:
    use: 'otobo-znuny:OTOBOZnunyTicketSystemService'
    params:
      base_url: "{{ get_env('OTOBO_URL') }}"
      username: "{{ get_env('OTOBO_USER') }}"
      password: "{{ get_env('OTOBO_PASSWORD') }}"

  classifier:
    use: 'hf-local:HuggingFaceLocalTextClassificationService'
    params:
      model_name: 'distilbert-base-uncased-finetuned-sst-2-english'

# Use plugin pipes in your pipeline
orchestrator:
  runners:
    - run:
        id: fetch_tickets
        use: 'base:FetchTicketsPipe'
        injects:
          ticket_system: 'ticket_system'
        params:
          ticket_search_criteria:
            queue_name: 'Support'
            state: 'new'
            limit: 25

    - run:
        id: classify_tickets
        use: 'hf-local:TextClassificationPipe'
        injects:
          classifier: 'classifier'
        params:
          tickets: "{{ get_pipe_result('fetch_tickets', 'fetched_tickets') }}"

    - run:
        id: update_tickets
        use: 'base:UpdateTicketPipe'
        injects:
          ticket_system: 'ticket_system'
        params:
          ticket_id: '{{ ticket.id }}'
          queue: '{{ classification.label }}'

Available Plugins

Base Plugin

Package: otai-base (installed automatically with Open Ticket AI)

The Base plugin provides core functionality that works with any ticket system:

Pipes:

• base:FetchTicketsPipe - Fetch tickets from your ticket system
• base:UpdateTicketPipe - Update ticket fields (queue, priority, state, etc.)
• base:AddNotePipe - Add notes or comments to tickets
• base:ExpressionPipe - Evaluate dynamic expressions and conditional logic

Example:

orchestrator:
  runners:
    - run:
        id: fetch_new_tickets
        use: 'base:FetchTicketsPipe'
        injects:
          ticket_system: 'my_ticket_system'
        params:
          ticket_search_criteria:
            state: 'new'
            limit: 50

OTOBO/Znuny Plugin

Package: otai-otobo-znuny

Connect to OTOBO, Znuny, and OTRS ticket systems.

Installation:

uv add otai-otobo-znuny

What it provides:

• Full integration with OTOBO, Znuny, and OTRS
• Ticket fetching, updating, and note management
• Queue, priority, and state management

Example Configuration:

services:
  otobo:
    use: 'otobo-znuny:OTOBOZnunyTicketSystemService'
    params:
      base_url: 'https://helpdesk.example.com'
      username: "{{ get_env('OTOBO_USER') }}"
      password: "{{ get_env('OTOBO_PASSWORD') }}"

orchestrator:
  runners:
    - run:
        id: fetch
        use: 'base:FetchTicketsPipe'
        injects:
          ticket_system: 'otobo'
        params:
          ticket_search_criteria:
            queue_name: 'IT Support'
            state: 'new'

HuggingFace Local Plugin

Package: otai-hf-local

Run machine learning classification models locally using HuggingFace transformers.

Installation:

uv add otai-hf-local

What it provides:

• Local ML model inference (no external API calls)
• Text classification for ticket content
• Support for any HuggingFace text classification model

Example Configuration:

services:
  ml_classifier:
    use: 'hf-local:HuggingFaceLocalTextClassificationService'
    params:
      model_name: 'distilbert-base-uncased-finetuned-sst-2-english'
      # Or use your custom fine-tuned model
      # model_name: "my-org/custom-ticket-classifier"

orchestrator:
  runners:
    - run:
        id: classify
        use: 'hf-local:TextClassificationPipe'
        injects:
          classifier: 'ml_classifier'
        params:
          text: '{{ ticket.title }} {{ ticket.body }}'

Benefits of the Plugin System

Install Only What You Need:

• Don’t use OTOBO? Don’t install the OTOBO plugin
• Running classification in the cloud? Skip the HuggingFace plugin
• Keeps your installation lightweight

Mix and Match:

• Use OTOBO plugin with HuggingFace models
• Combine multiple ticket systems in one setup
• Add community plugins as they become available

Independent Updates:

• Update plugins without updating core
• Get new features faster
• Roll back individual plugins if needed

Plugin Naming Convention

When you use plugins in configuration, the format is:

plugin-name:ComponentName

Examples:

• base:FetchTicketsPipe - FetchTicketsPipe from the base plugin
• otobo-znuny:OTOBOZnunyTicketSystemService - Service from the otobo-znuny plugin
• hf-local:TextClassificationPipe - Pipe from the hf-local plugin

The plugin name is derived from the package name by removing otai- prefix.

For Developers

Want to create your own plugin? Whether free or commercial, you have complete freedom!

See the Plugin Development Guide for:

• How to build custom plugins
• Publishing to PyPI
• Monetization strategies for commercial plugins
• Getting your plugin listed

Related Documentation

• Plugin Development
• Dependency Injection
• Pipe System
• Configuration