shinka-run

Execute batch evolution tasks via the ShinkaEvolve CLI with explicit run arguments.

1.3k|268|Updated Sep 17, 2025

Search Tags:#automation #cli #evolution #experimentation #shinka #task-dir #batch-run

Ready to run

npx skills add https://github.com/SakanaAI/ShinkaEvolve --skill shinka-run

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: shinka-run
Source: https://github.com/SakanaAI/ShinkaEvolve/tree/main/skills/shinka-run
Command: npx skills add https://github.com/SakanaAI/ShinkaEvolve --skill shinka-run

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

ShinkaRun enables agents to trigger and manage ShinkaEvolve task batches efficiently from a task directory, without manual setup.

Core Features & Use Cases

Run batches of evolutions from the evaluate.py script and initial.* task files within a single task directory.
Configure and launch runs deterministically using explicit CLI arguments like --results_dir, --num_generations, and per-run overrides.
Improve automation workflows for agents, experiments, and pipelines by quickly spinning up controlled evolution tasks.

Quick Start

Invoke shinka_run from the task directory to start a batch with a specific results directory and generation settings.

Frequently Asked Questions about shinka-run

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I run batch evolution tasks from a task directory?▼

Batch evolution tasks are executed by invoking the shinka_run CLI from a task directory containing evaluate.py and initial.* files, specifying --results_dir, --num_generations, and per-run overrides to generate and archive results.

What files are required in a task directory to launch ShinkaEvolve runs?▼

Launching ShinkaEvolve runs requires a valid task directory containing an evaluate.py script and initial.* task files, plus the shinka_run CLI available in the environment to execute the batch.

Can I configure specific results directories and generation counts for automated evolution experiments?▼

You can configure automated evolution experiments deterministically using explicit CLI arguments including --results_dir for output paths, --num_generations for counts, and per-run overrides for customized execution.

Does ShinkaEvolve batch execution work with automation agents and orchestration scripts?▼

ShinkaEvolve batch execution applies directly to automation agents or orchestration scripts that need to launch controlled generations, driving the CLI to generate, evaluate, and archive evolution results.

What are the limitations of running ShinkaEvolve batches from the command line?▼

Running ShinkaEvolve batches requires the shinka_run CLI to be available in the environment and a properly structured task directory, meaning execution fails without these explicit dependencies and file structures.