Selects code, generates tests with LLM and baseline, runs them with coverage, and reports results.
QualityFlow is an AI-powered test generation pipeline that demonstrates how to
build production-ready workflows for automated unit test generation using
Large Language Models (LLMs) with ZenML. It focuses on generating tests,
comparing against a baseline, and producing real coverage analysis.
Integrates LLM providers (OpenAI/Anthropic) for intelligent unit test generation
Selects target source files via configurable strategies with max_files limits
Generates baseline heuristic tests to compare with LLM-generated tests
Executes tests and produces coverage metrics and detailed reports
Tracks token usage and estimated costs via run metadata
A single Generate & Evaluate pipeline orchestrates: input selection, source fetching,
code analysis, LLM test generation, baseline test generation, test execution with
coverage, and comparative reporting
Uses ZenML Path artifacts to support remote/containerized execution
Configurable via YAML files to control speed, providers, and test limits
The architecture connects a streamlined test-generation pipeline to LLM providers and
test execution backends, producing coverage reports and markdown summaries that are
versioned and tracked in ZenML.
