yujunzhixue/studyanalysis-skills

Overview

This skill deeply ingests links, documents, or code and produces tutor-grade teaching notes that take learners from zero to mastery. It includes a Truth Anchoring Protocol that detects hallucinations and outdated statements, and it outputs both Markdown and HTML packaged for delivery.

How this skill works

The skill runs an ingestion script to clean and normalize source content into Markdown, then extracts key factual claims and runs time-aware web audits to validate them. After calibration, it loads a mentor persona prompt and generates structured, multi-modal teaching materials following mandated narrative constructs, presets, and output rules. Final QA checks enforce format, assessment items, cognitive maps, and domain-specific presentation styles.

When to use it

• User explicitly asks to learn, analyze, or deeply parse a URL, document, or codebase.
• User uploads multimodal content (PDF, Word, Markdown, images with text, or multiple links).
• User requests deep code analysis: execution flow, APIs, or architecture explanations.
• User asks for simplified explanations or stepwise learning from novice to expert.
• When accuracy and recency matter and claims must be validated against 2026-era sources.

Best practices

• Always run the ingestion command to produce a cleaned raw_content.txt before analysis.
• Include the current year in verification queries to catch stale or deprecated facts.
• Preserve and follow the system prompt’s mandated structure and labels—do not invent or omit mandatory modules.
• Generate both Markdown and HTML outputs and place them in a dated knowledge folder with clear filenames.
• Perform the full QA checklist (script tags, domain-specific headings, mermaid map, 5–8 self-tests) before delivery.

Example use cases

• Turn a research PDF into a stepwise learning guide with verified claims, concept maps, and practice questions.
• Ingest a GitHub repo to produce annotated tutor notes explaining architecture, key functions, and security caveats.
• Convert a long-form blog series into a structured course module with calibration notes on contested or outdated points.
• Analyze images of diagrams or slides (OCR) and integrate the cleaned text into a cohesive study guide.
• Audit vendor documentation for accuracy and produce a corrected, teachable summary for onboarding teams.