shinka-inspect

Extract top Shinka programs into a compact Markdown bundle.

Updated May 20, 2026
One-click install
npx skills add https://github.com/nhatnguyen1122/shinka --skill shinka-inspect-nhatnguyen1122
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: shinka-inspect
Source: https://github.com/nhatnguyen1122/shinka/tree/main/skills/shinka-inspect
Command: npx skills add https://github.com/nhatnguyen1122/shinka --skill shinka-inspect-nhatnguyen1122

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pandas, and includes scripts (resource) components.

What problem does it solve?

Loads top performing Shinka programs from a completed run into an agent-friendly context, enabling rapid iteration planning.

Core Features & Use Cases

  • Extracts top programs ranked by the combined score and formats them into a compact, loadable Markdown bundle suitable for agent contexts.
  • Supports optional filtering by generation and includes program metadata and code snippets for quick review.
  • Use Case: After a Shinka run finishes, generate a context bundle to guide the next mutation planning cycle in an autonomous agent.

Quick Start

Run the inspect script with a Shinka results directory to generate a compact Markdown context bundle of top programs for the next mutation planning step.

Frequently Asked Questions about shinka-inspect

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

FAQPage Schema
How do I load top performing programs from a Shinka run into an agent context?

To extract top programs from a Shinka run, execute the inspect script against a valid results directory to rank programs by combined score and output a compact Markdown bundle with program metadata and code snippets for agent planning.

What is a Shinka context bundle and when do I need it for agent planning?

A Shinka context bundle is a structured Markdown file containing top-ranked programs and their metadata from a completed run. It is needed when preparing an autonomous agent for the next mutation planning cycle after a run finishes.

Does generating a Shinka program context bundle require pandas and a SQLite database?

Yes, generating the bundle requires a Python environment with pandas to read and rank data, and a populated SQLite database within a valid Shinka results directory to extract program metadata and code snippets.

Can I filter Shinka programs by generation when building a Markdown context bundle?

Yes, the inspect script supports optional filtering by generation, allowing you to narrow the Markdown context bundle to specific generations of top-ranked programs from the completed run.

What are the limitations of using the shinka-inspect script for program ranking?

The inspect script is limited to runs with a valid results directory and a populated SQLite database, cannot process incomplete runs, and outputs only a Markdown bundle without direct code execution capabilities.