# SakanaAI/ShinkaEvolve

ShinkaEvolve: Towards Open-Ended and Sample-Efficient Program Evolution 🧬

Repository: https://github.com/SakanaAI/ShinkaEvolve
Canonical: https://ross.abutalabs.com/products/shinkaevolve
Homepage: https://sakanaai.github.io/ShinkaEvolve/
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-08-21T13:07:37+00:00

## Health v2 (maintenance only)
Score: 79/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 86, longevity 25
- inputs: {"age_days": 350, "days_push": 12, "days_rel": 92, "gap_med": 11.5, "n_releases_24m": 7}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1355, forks 272 (observed 2026-08-28T04:04:29.226372+00:00)

## What it is
ShinkaEvolve is a Python framework that combines LLMs with evolutionary algorithms to automatically evolve and improve scientific code. It maintains a population of programs with LLMs acting as mutation operators, supporting async evaluation, archive-based search, and local or SLURM cluster execution.

## Use cases
- evolve scientific code with llm-guided mutation
- run open-ended program evolution experiments
- optimize algorithms like circle packing automatically
- use llms as mutation operators for evolutionary search
- scale code evolution runs on a slurm cluster
- integrate program evolution into coding agents

## When to choose
- you want sample-efficient, LLM-driven evolutionary optimization of code
- you need reproducible, resumable evolution runs with a WebUI for inspection
- you want to run evolution locally or on clusters with Hydra/SLURM support

## When to avoid
- you need a general-purpose AutoML or hyperparameter tuning tool
- your task doesn't involve evolving or generating code
- you can't access LLM APIs or local models for mutation

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, llm-inference, machine-learning, cli, workflow-automation
- domain: artificial-intelligence, large-language-models, developer-tools, data-science
- platform: python, cli
- tags: evolutionary-algorithms, llm-guided-search, program-synthesis, scientific-discovery, code-evolution, hydra, slurm, ai-agents, linux, macos, docker

## Member repositories
- SakanaAI/ShinkaEvolve (main) score 79

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:29.226372+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:41:57.200901+00:00, confidence not recorded.
  - readme: https://github.com/SakanaAI/ShinkaEvolve (fetched 2026-08-28T04:04:29.226372+00:00, sha bfab56998db5)
  - homepage: https://sakanaai.github.io/ShinkaEvolve/ (fetched 2026-08-29T12:00:14.657437+00:00, sha 2d508843cc4d)
- Data as of 2026-08-30T08:39:29.467469+00:00.