darwinian-evolver

You can automatically optimize LLM prompts and small code blocks by running an evolutionary loop that proposes improvements and selects high-performing variants. Darwinian Evolver drives this exploration using a pluggable evaluator and mutator chain across multiple iterations.

Evolutionary optimization for prompt engineering is an automated process where an LLM-driven loop mutates candidate prompts and scores them against custom evaluators. This mechanism enables iterative improvement by automatically selecting the highest-performing prompt variants.

Yes, the evolutionary loop works with OpenRouter and OpenAI providers. The framework requires Python 3 and integrates with these compatible LLM providers to drive the mutator chain and generate optimized artifacts.

You can tune regex and SQL queries using an evolutionary loop by providing an initial artifact as input. The framework applies pluggable mutators to propose variations and evaluators to score them, automatically selecting the most effective regex or SQL variants over multiple iterations.

Yes, you can customize the scoring and mutation strategies for code optimization. The framework supports pluggable evaluators and mutators, allowing you to tailor custom scoring functions and mutation strategies to fit your specific optimization scenarios.

Limitations of using an LLM-driven evolutionary loop include the requirement for upstream tooling dependencies, Python 3, and compatible LLM provider access. It is designed for small code blocks and prompts, meaning large-scale software optimization is outside its intended scope.