skills.rest
tan-yong-sheng
tan-yong-shengCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

shinka-inspect

Rank Shinka programs by combined_score and emit a Markdown context bundle.

1|1|Updated Mar 5, 2026
Search Tags:#markdown#analysis#ranking#context-bundle#programs#shinka
One-click install
npx skills add https://github.com/tan-yong-sheng/GrowChat --skill shinka-inspect-tan-yong-sheng
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: shinka-inspect
Source: https://github.com/tan-yong-sheng/GrowChat/tree/main/.claude/skills/shinka-inspect
Command: npx skills add https://github.com/tan-yong-sheng/GrowChat --skill shinka-inspect-tan-yong-sheng

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Extracting and organizing the best-performing Shinka programs from a run can be tedious and manual. This skill surfaces top candidates, bundles them into a single Markdown artifact, and makes it easy to load context for planning iterations.

Core Features & Use Cases

  • Ranks programs by combined_score to surface top candidates, preferring correct results when available, with a fallback to top-scoring programs otherwise.
  • Generates one Markdown bundle that includes run metadata, a ranking table, per-program details with code blocks, and optional feedback for iteration planning.
  • Works with Shinka run artifacts in a SQLite database, enabling streamlined handoff to subsequent tasks and experiments.

Quick Start

Run the shinka-inspect skill against your Shinka run artifacts to generate the planning bundle.

Frequently Asked Questions about shinka-inspect

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

FAQPage Schema
How do I extract the top-performing Shinka programs from a run?▼

To extract top-performing Shinka programs, the skill processes run artifacts and ranks records by combined_score. It prefers correct rows and falls back to top-k overall, generating a Markdown bundle for planning.

What is the best way to organize Shinka run artifacts into a planning bundle?▼

The best way to organize Shinka run artifacts is to rank programs by combined_score and render them into a compact Markdown context bundle. This includes run metadata, a ranking table, and per-program details.

How does ranking by combined_score work when processing Shinka SQLite databases?▼

Ranking by combined_score loads program records via pandas from the Shinka SQLite database. It selects the top-k programs, applying a correct-first preference before falling back to top-scoring overall results.

Does this tool require pandas to generate the Markdown analysis of Shinka runs?▼

Yes, generating the Markdown analysis of Shinka runs requires pandas as a dependency. It uses pandas to load program records from the SQLite database before ranking and formatting the context bundle.

What happens if there are no correct rows in the Shinka results directory?▼

If there are no correct rows in the Shinka results directory, the ranking mechanism falls back to selecting the top-k overall programs by combined_score. This ensures the Markdown bundle still contains top candidates.

Can I include iteration feedback when exporting top Shinka programs to Markdown?▼

Yes, you can include optional feedback for iteration planning when exporting top Shinka programs to Markdown. The generated artifact contains per-program details, code blocks, and this feedback section.

Related Skills

Explore More Data & Analytics Skills→
nhatnguyen1122nhatnguyen1122

shinka-inspect

Extract top Shinka programs into a compact Markdown bundle.

Community
Intermediate
EmaRimoldiEmaRimoldi

hypothesis-revision

Revises research hypotheses based on failure diagnosis and tracks decisions.

Community
Advanced
2233admin2233admin

shinka-inspect

Extract top-ranked Shinka programs from a run database into a Markdown context bundle.

Community
Intermediate
EverettFieldEverettField
2

cloudbase-document-database-in-wechat-miniprogram

Manage CloudBase document databases from WeChat Mini Programs via wx.cloud.database().

Community
Intermediate