structurecc 2.1.0 → 3.0.1

package/.claude-plugin/plugin.json

@@ -0,0 +1,8 @@
+{
+  "name": "structurecc",
+  "version": "3.0.0",
+  "description": "Extract structured data from documents using Claude vision and parallel Task agents",
+  "author": {
+    "name": "UTMB Diagnostic Center"
+  }
+}

package/README.md

@@ -1,106 +1,196 @@
-<h1 align="center">STRUCTURE</h1>
+# structurecc
 
-<p align="center">
-<strong>Extract structured data from PDFs, Word docs, and images using Claude Code.</strong>
-</p>
+Document Structure Extraction for Claude Code
 
-<p align="center">
-<a href="https://www.npmjs.com/package/structurecc"><img src="https://img.shields.io/npm/v/structurecc.svg" alt="npm version"></a>
-<a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/License-MIT-yellow.svg" alt="License: MIT"></a>
-</p>
+Extract structured data from PDFs, Word documents, and images using Claude's native vision capabilities and parallel Task agents.
 
-<p align="center">
-<img src="assets/terminal.png" alt="structurecc" width="550">
-</p>
+## Installation
 
----
-
-## Requirements
-
-- **Node.js** - [nodejs.org](https://nodejs.org/)
-- **Claude Code** - Requires API key or Pro/Max subscription
+```bash
+npx structurecc
+```
 
----
+This installs the plugin to `~/.claude/plugins/structurecc/`.
 
-## Install
+## Usage
 
-### Step 1: Install Claude Code
+### Single Document
 
 ```bash
-npm install -g @anthropic-ai/claude-code
+/structure document.pdf
+/structure lab_image.png
+/structure report.docx
 ```
 
-<p align="center">
-<img src="assets/screenshots/step0.png" alt="Install Claude Code" width="550">
-</p>
-
-### Step 2: Install structurecc
+### Batch Processing
 
 ```bash
-npx structurecc
+/structure:batch ./documents/
+/structure:batch ./patient_files/ --output ./extracted/
 ```
 
-<p align="center">
-<img src="assets/screenshots/step1.png" alt="Install structurecc" width="420">
-</p>
+## Supported Formats
 
-### Step 3: Start Claude Code
+| Format | Extension | Notes |
+|--------|-----------|-------|
+| PDF | `.pdf` | Multi-page supported, chunked for large documents |
+| Word | `.docx`, `.doc` | Text and embedded images extracted |
+| Images | `.png`, `.jpg`, `.jpeg`, `.tiff`, `.bmp` | Single-page extraction |
 
-Navigate to your document folder and run:
+## Output
+
+For each document, structurecc generates:
 
-```bash
-cd ~/Desktop/documents
-claude
+```
+document_extracted/
+├── chunks/              # Individual chunk extractions (for debugging)
+├── structure.json       # Complete structured extraction
+└── STRUCTURE.md         # Human-readable markdown summary
 ```
 
-<p align="center">
-<img src="assets/screenshots/step3a.png" alt="Start Claude Code" width="460">
-</p>
+### structure.json
+
+```json
+{
+  "source": "/path/to/document.pdf",
+  "extracted": "2026-01-30T14:30:22Z",
+  "pages": [
+    {
+      "page": 1,
+      "elements": [
+        {
+          "id": "element_1",
+          "type": "table",
+          "title": "Table 1. Lab Results",
+          "data": {
+            "headers": ["Test", "Result", "Units", "Reference"],
+            "rows": [
+              ["Glucose", "126", "mg/dL", "70-100"]
+            ]
+          },
+          "confidence": 0.98
+        }
+      ]
+    }
+  ],
+  "summary": {
+    "total_pages": 5,
+    "tables": 3,
+    "figures": 4,
+    "equations": 1,
+    "average_confidence": 0.94
+  }
+}
+```
 
-### Step 4: Run structure
+## Architecture
 
-Inside Claude Code:
+structurecc uses a chunk-based parallel processing approach:
+
+1. **Document Analysis** - Determine page count and split into chunks (5 pages each)
+2. **Parallel Extraction** - Launch one Task agent per chunk for parallel processing
+3. **Chunk Merge** - Combine chunk results with page offset correction
+4. **Output Generation** - Create JSON and Markdown outputs
 
 ```
-/structure document.pdf
+Document (20 pages)
+       │
+       ├── Chunk 1 (Pages 1-5)  → Agent 1
+       ├── Chunk 2 (Pages 6-10) → Agent 2
+       ├── Chunk 3 (Pages 11-15)→ Agent 3
+       └── Chunk 4 (Pages 16-20)→ Agent 4
+               │
+               ▼
+         Merged Output
 ```
 
-<p align="center">
-<img src="assets/screenshots/step3.png" alt="Run /structure" width="520">
-</p>
+This approach:
+- Maximizes throughput via parallel processing
+- Preserves context within chunks (figures and captions stay together)
+- Uses Claude's native vision (no external APIs)
+- Each agent has 200K context for thorough extraction
 
-Supports **PDF**, **DOCX**, **PNG**, **JPG**, and **TIFF**.
+## Element Types
 
----
+### Tables
 
-## Output
+Extracted with:
+- Headers and all rows
+- Cell values with exact formatting
+- Flags (H, L, *, †)
+- Footnotes
+- Merged cell information
 
-```
-document_extracted/
-├── images/              # Extracted visuals
-├── elements/            # Markdown per element
-└── STRUCTURED.md        # Combined output
-```
+### Figures
 
----
+Supports various figure types:
+- **Charts/Graphs**: Line, bar, scatter, pie with data series and axes
+- **Scientific Images**: Western blots, gels, micrographs
+- **Diagrams**: Flowcharts, illustrations, photographs
 
-## Troubleshooting
+Each figure includes:
+- Title and caption
+- Data points (when visible)
+- Axis labels and ranges
+- Annotations and legends
 
-| Issue | Solution |
-|-------|----------|
-| `npm: command not found` | Install Node.js from [nodejs.org](https://nodejs.org/) |
-| `/structure: No such file` | Run `claude` first, then type `/structure` inside Claude Code |
-| No images found | PDF may be text-only with no embedded images |
+### Equations
 
----
+Extracted as:
+- LaTeX representation
+- Plain text fallback
+- Variable definitions
 
-## Uninstall
+### Text Blocks
 
-```bash
-npx structurecc --uninstall
-```
+Captured with:
+- Full content
+- Type (header, paragraph, caption, footnote)
+- Formatting information
+
+## Confidence Scores
+
+Every element includes a confidence score (0.0-1.0):
+
+| Score | Meaning |
+|-------|---------|
+| 0.95-1.00 | Crystal clear extraction |
+| 0.85-0.94 | Clear with minor uncertainty |
+| 0.70-0.84 | Readable but some ambiguity |
+| < 0.70 | Needs manual verification |
+
+Low confidence items are flagged in the output for review.
+
+## Use Cases
+
+- **Medical Lab Results**: Extract patient data from PDF reports
+- **Research Papers**: Structure tables and figures from publications
+- **Scientific Images**: Transcribe gel/blot data for documentation
+- **Patient Records**: Batch process document folders
+- **Data Digitization**: Convert scanned documents to structured data
+
+## Requirements
+
+- Claude Code CLI
+- No external dependencies (uses Claude's native capabilities)
+
+## How It Works
+
+structurecc leverages Claude's multimodal capabilities:
+
+1. **Claude Vision**: Reads PDFs and images natively without OCR
+2. **Parallel Agents**: Task tool spawns chunk agents for parallel processing
+3. **Structured Output**: JSON schema ensures consistent, parseable output
+4. **Markdown Summary**: Human-readable format for quick review
+
+No web searches, no external APIs, no Python dependencies. Just Claude + document = structured data.
+
+## Limitations
 
----
+- Very large documents (100+ pages) may require multiple runs
+- Handwritten content has lower accuracy than printed text
+- Low-resolution images may have reduced confidence scores
+- Complex nested tables may require manual verification
 
 ## License
 

package/commands/structure-batch.md

@@ -0,0 +1,276 @@
+---
+description: Batch extract structured data from multiple documents in a directory
+argument-hint: <directory>
+---
+
+# Batch Document Structure Extraction
+
+You are extracting structured data from multiple documents in a directory.
+
+## Input
+
+**Directory path:** $ARGUMENTS
+
+Parse arguments:
+- First argument: Directory path (required)
+- `--output <dir>`: Custom output directory (optional, defaults to source directory)
+- `--pattern <glob>`: File pattern to match (optional, defaults to all supported types)
+
+## Workflow
+
+### Step 1: Discover Documents
+
+Use Glob to find all supported documents in the directory:
+
+```
+Supported patterns:
+- *.pdf
+- *.docx
+- *.doc
+- *.png
+- *.jpg
+- *.jpeg
+- *.tiff
+- *.bmp
+```
+
+List all discovered documents:
+
+```
+┌──────────────────────────────────────────────────────────────────────┐
+│  BATCH EXTRACTION                                                    │
+├──────────────────────────────────────────────────────────────────────┤
+│                                                                      │
+│  Directory: {path}                                                   │
+│                                                                      │
+│  Documents Found: {count}                                            │
+│    1. document1.pdf (15 pages)                                       │
+│    2. lab_results.png (1 image)                                      │
+│    3. report.docx (8 pages)                                          │
+│    ...                                                               │
+│                                                                      │
+│  Estimated chunks: {total_chunks}                                    │
+│                                                                      │
+└──────────────────────────────────────────────────────────────────────┘
+```
+
+### Step 2: Create Batch Output Directory
+
+```
+{output_dir}/
+├── batch_summary.json       # Summary of all extractions
+├── batch_summary.md         # Human-readable summary
+├── document1_extracted/     # Per-document outputs
+│   ├── structure.json
+│   └── STRUCTURE.md
+├── lab_results_extracted/
+│   ├── structure.json
+│   └── STRUCTURE.md
+└── report_extracted/
+    ├── structure.json
+    └── STRUCTURE.md
+```
+
+### Step 3: Process Documents
+
+For each document, invoke the /structure command:
+
+**Option A: Sequential Processing (safer, easier to track)**
+Process one document at a time, reporting progress:
+
+```
+Processing document 1/3: document1.pdf
+  Launching 3 chunk agents...
+  Chunks complete. Merging...
+  Done. Tables: 5, Figures: 3
+
+Processing document 2/3: lab_results.png
+  Processing as single chunk...
+  Done. Tables: 1, Figures: 0
+
+Processing document 3/3: report.docx
+  Launching 2 chunk agents...
+  Chunks complete. Merging...
+  Done. Tables: 2, Figures: 4
+```
+
+**Option B: Parallel Documents (faster for many small docs)**
+For directories with many single-page documents (like images), launch multiple document extractions in parallel using Task agents.
+
+Choose based on:
+- <5 documents or large PDFs: Sequential
+- >5 small documents (images, short PDFs): Parallel
+
+### Step 4: Generate Batch Summary
+
+**batch_summary.json:**
+
+```json
+{
+  "batch": {
+    "directory": "/path/to/documents",
+    "processed": "2026-01-30T14:30:22Z",
+    "document_count": 3,
+    "total_pages": 24,
+    "total_elements": 45
+  },
+  "documents": [
+    {
+      "filename": "document1.pdf",
+      "pages": 15,
+      "tables": 5,
+      "figures": 3,
+      "equations": 1,
+      "average_confidence": 0.94,
+      "output": "document1_extracted/structure.json"
+    },
+    {
+      "filename": "lab_results.png",
+      "pages": 1,
+      "tables": 1,
+      "figures": 0,
+      "equations": 0,
+      "average_confidence": 0.97,
+      "output": "lab_results_extracted/structure.json"
+    },
+    {
+      "filename": "report.docx",
+      "pages": 8,
+      "tables": 2,
+      "figures": 4,
+      "equations": 0,
+      "average_confidence": 0.91,
+      "output": "report_extracted/structure.json"
+    }
+  ],
+  "summary": {
+    "total_tables": 8,
+    "total_figures": 7,
+    "total_equations": 1,
+    "overall_confidence": 0.94,
+    "low_confidence_items": 2
+  }
+}
+```
+
+**batch_summary.md:**
+
+```markdown
+# Batch Extraction Summary
+
+**Directory:** {path}
+**Processed:** {timestamp}
+**Documents:** {count}
+
+---
+
+## Overview
+
+| Metric | Count |
+|--------|-------|
+| Total Documents | 3 |
+| Total Pages | 24 |
+| Total Tables | 8 |
+| Total Figures | 7 |
+| Total Equations | 1 |
+| Overall Confidence | 94% |
+
+---
+
+## Document Details
+
+### 1. document1.pdf
+
+- **Pages:** 15
+- **Tables:** 5
+- **Figures:** 3
+- **Confidence:** 94%
+- **Output:** [document1_extracted/](document1_extracted/)
+
+### 2. lab_results.png
+
+- **Pages:** 1
+- **Tables:** 1
+- **Figures:** 0
+- **Confidence:** 97%
+- **Output:** [lab_results_extracted/](lab_results_extracted/)
+
+### 3. report.docx
+
+- **Pages:** 8
+- **Tables:** 2
+- **Figures:** 4
+- **Confidence:** 91%
+- **Output:** [report_extracted/](report_extracted/)
+
+---
+
+## Low Confidence Items
+
+Items with confidence < 80% requiring review:
+
+1. **document1.pdf, Page 7, Table 3** (72%)
+   - Reason: Partially obscured by watermark
+
+2. **report.docx, Page 5, Figure 2** (68%)
+   - Reason: Low resolution image
+
+---
+
+## Files Generated
+
+```
+{output_dir}/
+├── batch_summary.json
+├── batch_summary.md
+├── document1_extracted/
+│   ├── structure.json
+│   └── STRUCTURE.md
+├── lab_results_extracted/
+│   ├── structure.json
+│   └── STRUCTURE.md
+└── report_extracted/
+    ├── structure.json
+    └── STRUCTURE.md
+```
+```
+
+### Step 5: Display Completion
+
+```
+┌──────────────────────────────────────────────────────────────────────┐
+│  BATCH EXTRACTION COMPLETE                                           │
+├──────────────────────────────────────────────────────────────────────┤
+│                                                                      │
+│  Documents Processed: {count}                                        │
+│  Total Pages: {pages}                                                │
+│  Total Elements: {elements}                                          │
+│                                                                      │
+│  Summary:                                                            │
+│    Tables: {count}                                                   │
+│    Figures: {count}                                                  │
+│    Equations: {count}                                                │
+│                                                                      │
+│  Overall Confidence: {score}%                                        │
+│  Items Needing Review: {count}                                       │
+│                                                                      │
+│  Output:                                                             │
+│    {output_dir}/batch_summary.json                                   │
+│    {output_dir}/batch_summary.md                                     │
+│                                                                      │
+└──────────────────────────────────────────────────────────────────────┘
+```
+
+## Error Handling
+
+- If directory doesn't exist: Report error
+- If no supported documents found: Report and suggest checking patterns
+- If individual document fails: Log error, continue with others, report in summary
+- If output directory cannot be created: Report error
+
+## Notes
+
+- Each document is processed independently using /structure
+- Batch processing is ideal for patient files, lab results, research papers
+- Low confidence items are aggregated for easy review
+- All outputs are self-contained and can be accessed individually