shinka-convert

To convert an existing codebase into an evolvable task, snapshot your current working directory to generate a sidecar task directory containing initial.py, evaluate.py, run_evo.py, and shinka.yaml. This scaffolds a minimal runnable subset of your project for evolution.

A Shinka task directory is an isolated sidecar folder containing evaluation and runner artifacts for the ShinkaEvolve pipeline. You need it when you want to optimize legacy scripts or repositories through evolution instead of building a task from scratch.

Yes, you can use Shinka to optimize legacy scripts without starting from scratch by snapshotting your current codebase. The conversion process creates a sidecar task directory with initial artifacts, enabling reproducible evolution of your existing project.

Snapshotting a project for ShinkaEvolve generates evaluate.py, run_evo.py, shinka.yaml, initial.py, and a default shinka_task directory. These files provide a stable evaluation entry point and configuration for the evolution pipeline.

Yes, you need Shinka and PyYAML installed as dependencies to scaffold a task directory. These packages enable the conversion script to snapshot code and generate the required YAML configuration and Python evaluation artifacts.

A limitation of converting a codebase snapshot for evolution is that it only captures a minimal runnable subset of your project. Complex repositories may require manual adjustments to the generated evaluate.py and initial.py to ensure the evaluation pipeline runs correctly.