shinka-convert

To convert an existing codebase into an evolvable task, snapshot the relevant code, add evolve blocks, and generate evaluate.py plus runner/config files. This isolates evolvable regions for safe automated experimentation while preserving the original project.

Yes, you can optimize existing code with Shinka by snapshotting the minimal runnable codebase and generating an evolvable task scaffold. This avoids creating a new task from a natural-language description and allows direct automated evolution of your current project.

The best way to automate codebase evolution for an existing project is generating a ShinkaEvolve task directory that produces evaluate.py and Shinka runner/config files. This enables the Shinka pipeline to safely evaluate and evolve isolated regions of your code.

No, you do not need to modify your original project files to generate a Shinka task. The conversion process preserves the original code while producing a small, isolated evolution task directory for experimentation.

When snapshotting an existing project for Shinka evolution, the process generates an evolvable task scaffold containing evaluate.py and Shinka runner/config files. These files enable automated evaluation within the Shinka pipeline.

You should use automated codebase conversion instead of natural-language task generation when you want to optimize an existing project. It snapshots your current code to isolate evolvable regions, whereas natural-language generation creates brand-new tasks from descriptions.