Letta Pipeline for Open WebUI

A pipeline integration for Open WebUI that connects to the Letta AI API, providing streaming responses with event emitters and development tools.

Features

• 🔄 Streaming responses with real-time updates
• 🤔 Reasoning steps displayed in status field
• 📊 Usage statistics tracking
• 🛠️ Development mode with detailed logging
• 🔧 Configurable settings via UI
• 🔌 Tool integration with Open WebUI
• ✅ Comprehensive test suite

Message Types

The pipeline handles three types of messages:

1. assistant_message - Main response content
2. reasoning_message - Reasoning steps shown in status
3. usage_statistics - Token usage and performance stats

Status Messages

Status updates use emojis for better visibility:

• 🔄 Processing request...
• 🤔 Reasoning steps
• ✓ Response complete
• ⚠️ Error messages

Installation

1. Clone this repository:

git clone https://github.com/oculairmedia/letta-pipeline.git
cd letta-pipeline

2. Install dependencies:

pip install -r requirements.txt

3. Set up environment variables:

export LETTA_BASE_URL="https://letta2.oculair.ca"
export LETTA_AGENT_ID="your-agent-id"
export LETTA_PASSWORD="your-password"

Usage

1. Import and initialize the pipeline:

from letta import Pipe

pipe = Pipe()

2. Use the pipeline in Open WebUI:

# The pipeline will be automatically registered with Open WebUI
# and will appear in the model selection dropdown

Development

1. Enable development mode:

pipe.valves.DEV_MODE = True
pipe.valves.LOG_RAW_CHUNKS = True
pipe.valves.LOG_PARSED_CHUNKS = True
pipe.valves.LOG_EVENTS = True
pipe.valves.SAVE_RESPONSES = True

2. Run tests:

python test_letta_dev.py

License

MIT License