analyze

Analyze repository code to answer user questions with evidence and confidence scores.

1|Updated Sep 22, 2025

Search Tags:#codebase #evidence #confidence #analyze #read-only #file-references #ranked-synthesis

Ready to run

npx skills add https://github.com/prthik/prathik-astro --skill analyze-prthik

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: analyze
Source: https://github.com/prthik/prathik-astro/tree/main/.codex/skills/analyze
Command: npx skills add https://github.com/prthik/prathik-astro --skill analyze-prthik

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Read-only, evidence-driven analysis of the repository to explain what the codebase implies about a user question.

Core Features & Use Cases

Read-only examination that traces behavior across modules.
Generates a ranked synthesis with explicit evidence, confidence, and file references.
Distinguishes evidence from inference to support reliable decision-making.

Quick Start

Run the analyze skill to produce a ranked, evidence-based synthesis with file references and explicit confidence.

Frequently Asked Questions about analyze

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

FAQPage Schema

How do I trace behavior across modules without modifying the codebase?▼

You can trace behavior across modules safely using read-only code analysis. This approach examines the repository to explain what the codebase implies about your question without altering any files.

What is evidence-driven codebase analysis?▼

Evidence-driven codebase analysis is a method that separates verified code evidence from inferred conclusions. It generates a ranked synthesis with explicit confidence scoring and file references to support reliable decision-making.

How do I assess the impact of a change before modifying my code?▼

Perform a read-only impact assessment to trace behavior across modules. This provides an architecture overview and causality insights, helping you understand implications before making any changes.

Can I get explicit confidence scores for cross-file code explanations?▼

Yes, cross-file code explanations can include explicit confidence scores. The analysis distinguishes direct evidence from inference, providing a ranked synthesis with file references for reliability.

When do I need an evidence-vs-inference separation for code analysis?▼

You need evidence-vs-inference separation when requiring reliable architecture insights or causality assessments. This distinction ensures your decision-making is based on verified codebase evidence rather than assumptions.

Does read-only code analysis work for exploring large codebases?▼

Read-only code analysis applies bounded, discriminating exploration to navigate large codebases. It traces behavior across modules to produce a ranked synthesis with file references and explicit confidence.