Mystery Analysis Skill for Claude Code

Evidence-based mystery and detective fiction analysis for Claude Code.

This skill is built for readers who want more than a plot summary. It helps an agent read extracted book text, analyze chapters for clues and anomalies, and maintain an Obsidian vault with structured notes for characters, locations, objects, and timeline tracking.

Highlights

• Works well for mystery novels and detective fiction
• Supports EPUB-based workflows through agent-run extraction
• Produces Obsidian-ready vaults instead of one-off answers
• Tracks clues, anomalies, foreshadowing, and cross-chapter consistency
• Organizes entities into Characters/, Locations/, Objects/, and Timeline/

Install

Install from this repository:

git clone https://github.com/The-time-forever/mysterycode-skill.git
/plugin add /path/to/mysterycode-skill

Or copy it into the Claude skills directory:

cp -r mysterycode-skill ~/.claude/skills/

On Windows, copy the folder to C:\Users\<you>\.claude\skills\.

Verify

claude list-skills

Look for mystery-analysis.

What It Does

• Read mystery novels from EPUB-derived text and other text formats
• Analyze chapters for details, anomalies, foreshadowing, and insights
• Check consistency across chapters and timelines
• Generate relationship-oriented reading notes
• Build and maintain an Obsidian vault with structured folders and linked pages

Typical Use

You do not need to run the Python scripts manually in normal use. The agent should call them when needed.

Example prompts:

Read this EPUB, analyze chapter 1, and create an Obsidian vault in ./mystery-vault-book

Continue analyzing chapter 2 and update the existing Obsidian vault

Check this mystery vault for missing links, empty pages, and stale indexes

Obsidian Output

This skill is designed to produce and maintain an Obsidian vault, not just isolated chapter summaries.

Typical vault structure:

vault/
├── Books/[书名]/
│   ├── 00-书籍信息.md
│   ├── 01-第一章分析.md
│   └── ...
├── Analysis/
├── Characters/
├── Locations/
├── Objects/
├── Timeline/
└── Home.md

The vault is meant to stay navigable as it grows:

• Books/ stores chapter analysis
• Analysis/ stores cross-chapter clue tracking
• Characters/, Locations/, and Objects/ store entity pages
• Timeline/ stores chronology and flow notes
• Home.md acts as the vault entry page

How It Works

1. The agent extracts readable chapter text from EPUB when needed.
2. The agent analyzes selected chapters using evidence-first close reading.
3. The agent writes or updates the Obsidian vault.
4. The agent synchronizes chapter indexes, clue tracking, and entity pages.
5. The agent validates the vault to catch missing links, empty pages, and stale navigation.

Behavior Notes

• EPUB use is intentionally two-step: extract first, then analyze.
• EPUB extraction supports zh, en, and --language auto.
• Extracted notes are reference context, not primary plot evidence.
• Vault output location is user-controlled.
• If extraction fails, the skill should stop instead of guessing from prior knowledge.
• Failed extraction should be retried through controlled parameters such as --language, --include-item, --exclude-item, or --disable-frontmatter-filter.
• Chapter updates should also update book info, clue tracking, recent-analysis links, and minimal entity pages.

Repository Layout

mysterycode-skill/
├── SKILL.md
├── skill.json
├── scripts/
│   ├── extract_epub.py
│   └── validate_vault.py
├── templates/
│   ├── chapter-analysis-template.md
│   ├── character-card-template.md
│   ├── location-card-template.md
│   ├── object-card-template.md
│   ├── timeline-note-template.md
│   └── clue-tracking-template.md
└── docs/
    ├── README_CN.md
    ├── USAGE.md
    ├── prompt.md
    ├── examples.md
    ├── PROJECT_OVERVIEW.md
    └── CHANGELOG.md

More Docs

• Chinese overview: docs/README_CN.md
• Usage guide: docs/USAGE.md
• Prompt rules: docs/prompt.md
• Project overview: docs/PROJECT_OVERVIEW.md
• Examples: docs/examples.md
• Changelog: docs/CHANGELOG.md

License

MIT