llx

LLX Integration Example

This example shows how to integrate llx for intelligent code generation within pyqual quality pipelines.

Overview

LLX enhances pyqual by:

• Analyzing project metrics (files, lines, complexity, duplication)
• Selecting the optimal LLM model based on actual code metrics
• Generating targeted fixes using the selected model

Files

• pyqual-llx.yaml - Complete pipeline configuration with llx integration
• README.md - This file

Quick Start

1. Install dependencies:

   pip install llx[prellm] pyqual code2llm vallm

2. Copy the configuration:

   cp pyqual-llx.yaml ../../pyqual.yaml

3. Run the pipeline:

   cd ../..
   pyqual run

How It Works

1. Analyze: code2llm collects project metrics
2. Validate: vallm identifies issues and creates error report
3. Fix: llx fix reads errors, selects optimal model, generates fixes
4. Test: Run tests to verify fixes
5. Loop: Repeat until all quality gates pass

Model Selection

LLX automatically selects models based on project metrics:

Project Size	Files	Lines	Selected Model
Small	<3	<500	Free (Gemini 2.5 Pro)
Medium	3-10	500-5K	Cheap (Claude Haiku 4.5)
Large	10-50	5K-20K	Balanced (Claude Sonnet 4)
Very Large	50+	20K+	Premium (Claude Opus 4)

Customization

See the full documentation for:

• Custom model thresholds
• Advanced configuration options
• Error handling strategies
• Best practices