Literature Structuring Expert
Automatically decompose long documents (papers, reports, books) into a structured, AI-friendly knowledge base. Splits by chapter, generates machine-readable summaries, and builds a navigable index to overcome context limits.
When to Use This Skill
Use this skill when the user:
- Has a document that is too long for the AI's context window.
- Needs to perform cross-chapter analysis or get a high-level overview of a long text.
- Wants to build a reusable, queryable knowledge base from a PDF, Markdown, or text file.
- Asks: "How can I get my AI to read this whole book/paper?"
Quick Reference
| Situation |
Action |
| User provides a long document |
1. Analyze and split it into chapters.2. Generate a JSON summary for each chapter.3. Create a master index file. |
| User asks a high-level, cross-chapter question |
Provide the content of the MASTER_INDEX.md file to the AI. |
| User asks a detailed, chapter-specific question |
Provide the corresponding single file from the ./chapters/ directory to the AI. |
| Task completed |
Present the generated file tree and MASTER_INDEX.md preview to the user. |
Core Workflow
Phase 1: Intelligent Splitting
- Analyze Input: Receive the long document text or file path.
- Identify Structure: Automatically analyze the document to identify heading hierarchies (e.g.,
#, ##, 1., 1.1) to determine chapter boundaries. Prioritize user-specified splitting preferences.
- Execute Split: Split the document into independent plain-text files by chapter.
- Naming Convention:
{sequence_number}_{chapter_title}.md (e.g., 01_Introduction.md).
- Storage Location: All chapter files are saved in the
./chapters/ directory.
Phase 2: Summary Generation & Structuring
- Generate Summary per Chapter: For each file in
./chapters/, generate a corresponding JSON summary file.
- Structured Fields (JSON format):
{
"chapter_id": "Unique identifier matching the filename, e.g., 02_1",
"chapter_title": "Chapter Title",
"abstract": "Core summary of the chapter, 200-300 words.",
"keywords": ["Keyword1", "Keyword2", "Keyword3"],
"key_points": ["Key point one", "Key point two"],
"related_sections": ["IDs of other chapters strongly related to this one"]
}
- Storage Location: JSON summary files are saved in the
./summaries/ directory (e.g., 01_Introduction.summary.json).
Phase 3: Create Global Index
- Aggregate Information: Collect data from all JSON files in
./summaries/.
- Generate Index: Create a global index file,
MASTER_INDEX.md.
- Content: Lists all chapters' IDs, titles, a short abstract preview, and keywords in a Markdown list or table.
- Purpose: Provides a "bird's-eye view" for quick navigation and high-level Q&A.
Final Deliverables & File Structure
Upon completion, the following file tree is generated:
Project_Root/
├── chapters/ # 【Source Repository】Contains all split chapter texts (.md files)
│ ├── 01_Introduction.md
│ ├── 02_1_Experimental_Methods.md
│ └── ...
├── summaries/ # 【Summary Repository】Contains all structured JSON summaries
│ ├── 01_Introduction.summary.json
│ ├── 02_1_Experimental_Methods.summary.json
│ └── ...
└── MASTER_INDEX.md # 【Global Navigation】Core document summary index
Usage Instructions for the User
For Global, Cross-Chapter Queries (e.g., “What is the paper's main thesis?”):
- Provide the content of the
MASTER_INDEX.md file to the AI. This is token-efficient.
For Specific, In-Depth Queries Within a Chapter (e.g., “What were the parameters in the 'Methods' section?”):
- Provide the corresponding single chapter file from the
chapters/ directory to the AI for full context.
1---2name: large-document-reader3description: Intelligently splits long academic or technical documents into chapters, generates structured JSON summaries for each, and creates a file system with a global index. This enables efficient AI retrieval and analysis, perfectly solving context window limitations by enabling “overview via summaries, deep-dive on demand” workflows.4---5
6# Literature Structuring Expert
7
8Automatically decompose long documents (papers, reports, books) into a structured, AI-friendly knowledge base. Splits by chapter, generates machine-readable summaries, and builds a navigable index to overcome context limits.
9
10## When to Use This Skill
11
12Use this skill when the user:
13
14- Has a document that is too long for the AI's context window.
15- Needs to perform cross-chapter analysis or get a high-level overview of a long text.
16- Wants to build a reusable, queryable knowledge base from a PDF, Markdown, or text file.
17- Asks: "How can I get my AI to read this whole book/paper?"
18
19## Quick Reference
20
21| Situation | Action |
22|-----------|--------|
23| User provides a long document | 1. Analyze and split it into chapters.<br>2. Generate a JSON summary for each chapter.<br>3. Create a master index file. |
24| User asks a high-level, cross-chapter question | Provide the content of the `MASTER_INDEX.md` file to the AI. |
25| User asks a detailed, chapter-specific question | Provide the corresponding single file from the `./chapters/` directory to the AI. |
26| Task completed | Present the generated file tree and `MASTER_INDEX.md` preview to the user. |
27
28## Core Workflow
29
30### Phase 1: Intelligent Splitting
311. **Analyze Input**: Receive the long document text or file path.
322. **Identify Structure**: Automatically analyze the document to identify heading hierarchies (e.g., `#`, `##`, `1.`, `1.1`) to determine chapter boundaries. Prioritize user-specified splitting preferences.
333. **Execute Split**: Split the document into independent plain-text files by chapter.
34 * **Naming Convention**: `{sequence_number}_{chapter_title}.md` (e.g., `01_Introduction.md`).
35 * **Storage Location**: All chapter files are saved in the `./chapters/` directory.
36
37### Phase 2: Summary Generation & Structuring
381. **Generate Summary per Chapter**: For each file in `./chapters/`, generate a corresponding JSON summary file.
39 * **Structured Fields** (JSON format):
40 ```json
41 {
42 "chapter_id": "Unique identifier matching the filename, e.g., 02_1",
43 "chapter_title": "Chapter Title",
44 "abstract": "Core summary of the chapter, 200-300 words.",
45 "keywords": ["Keyword1", "Keyword2", "Keyword3"],
46 "key_points": ["Key point one", "Key point two"],
47 "related_sections": ["IDs of other chapters strongly related to this one"]
48 }
49 ```
50 * **Storage Location**: JSON summary files are saved in the `./summaries/` directory (e.g., `01_Introduction.summary.json`).
51
52### Phase 3: Create Global Index
531. **Aggregate Information**: Collect data from all JSON files in `./summaries/`.
542. **Generate Index**: Create a global index file, `MASTER_INDEX.md`.
55 * **Content**: Lists all chapters' IDs, titles, a short abstract preview, and keywords in a Markdown list or table.
56 * **Purpose**: Provides a "bird's-eye view" for quick navigation and high-level Q&A.
57
58## Final Deliverables & File Structure
59
60Upon completion, the following file tree is generated:
61```
62Project_Root/
63├── chapters/ # 【Source Repository】Contains all split chapter texts (.md files)
64│ ├── 01_Introduction.md
65│ ├── 02_1_Experimental_Methods.md
66│ └── ...
67├── summaries/ # 【Summary Repository】Contains all structured JSON summaries
68│ ├── 01_Introduction.summary.json
69│ ├── 02_1_Experimental_Methods.summary.json
70│ └── ...
71└── MASTER_INDEX.md # 【Global Navigation】Core document summary index
72```
73
74## Usage Instructions for the User
75
76**For Global, Cross-Chapter Queries** (e.g., “What is the paper's main thesis?”):
77* Provide the content of the `MASTER_INDEX.md` file to the AI. This is token-efficient.
78
79**For Specific, In-Depth Queries Within a Chapter** (e.g., “What were the parameters in the 'Methods' section?”):
80* Provide the corresponding single chapter file from the `chapters/` directory to the AI for full context.
modbender