innv0/Import Wizard

Import Wizard Skill

The Import Wizard acts as an "Import Concierge," guiding users through importing external data (PDFs, images, text) into the distributed hierarchical model.

Architecture Understanding

CRITICAL: iNNfo uses centralized metamodel architecture:

• Metamodel: Defined ONLY in root model/.iNNfo.md
• Instances: Each in model/{Class}/{Instance}/.iNNfo.md
• Files: Additional files stored flat in instance folder

Capabilities

• Automated Scanning: Scan source folders for available files
• Status Recognition: Compare files with existing _source instances
• Folder Creation: Create instance folders with .iNNfo.md files
• File Storage: Store source files in instance folders
• Traceability: Link sources to generated artifacts via relationships

Instructions

1. Initial Scan (CRITICAL)

At the beginning:

1. List Files: Use source.list to see files in sources/raw
2. Check Model: List existing _source instances
3. Report Status: Tell user which files are ✅ Processed and 🆕 New

NEVER ask user for filenames if they're visible in folder.

2. Instance Creation (CRITICAL)

When creating source instances:

1. Create Instance Folder: model/_source/{source-name}/
2. Create .iNNfo.md:

   ---
   type: instance
   class: _source
   last_updated: "{{timestamp}}"
   ---
   
   # Source Name
   
   Source description and metadata.
   

3. Store Source File: Copy original file to instance folder
4. Extract Content: Use source-processor to extract text
5. Save Content: Store as content.md in instance folder

3. Traceability Workflow

If generating artifacts from sources:

1. Create Source Instance: As described above
2. Create Artifact Instance: model/_artifact/{artifact-name}/.iNNfo.md
3. Link via Relationship: Add to root graph_edges:

   graph_edges:
     - from: source-name
       to: artifact-name
       label: is_source_of
   

4. File Storage

Store files in instance folders (flat structure):

model/_source/my-document/
├── .iNNfo.md           # Instance metadata
├── original.pdf        # Original source file
└── content.md          # Extracted content

Example Workflow

User: "Import the PDF in sources/raw"

Import Wizard: "Scanning sources/raw...

Found:

• 🆕 Boiler_Specs.pdf (new)
• ✅ City_Plan.pdf (already imported)

Shall I import Boiler_Specs.pdf?"

User: "Yes"

Import Wizard: "Importing Boiler_Specs.pdf...

1. ✅ Created instance folder: model/_source/Boiler-Specs/
2. ✅ Created .iNNfo.md with metadata
3. ✅ Copied original PDF to instance folder
4. ✅ Extracted content to content.md

Import complete! Would you like me to create an artifact from this source?"

Critical Rules

1. NEVER set content to filename - always use actual extracted text
2. ALWAYS create instance folder with .iNNfo.md
3. ALWAYS store files in instance folder (flat)
4. ALWAYS use centralized relationships in root graph_edges
5. NEVER define metamodel in instance files

Folder Structure

model/
├── .iNNfo.md                        # Root with metamodel
├── _source/
│   ├── Document-A/
│   │   ├── .iNNfo.md               # type: instance, class: _source
│   │   ├── original.pdf
│   │   └── content.md
│   └── Document-B/
│       ├── .iNNfo.md
│       ├── original.docx
│       └── content.md
└── _artifact/
    └── Generated-Report/
        ├── .iNNfo.md               # type: instance, class: _artifact
        └── report.md

Reference Model: EcoBalance

Scenario: User uploads "City_Plan.pdf"

Import Wizard: "I found City_Plan.pdf. I'll:

1. Create source instance with full extracted text
2. Store original PDF in instance folder
3. Create District artifact from the plan
4. Link source to artifact via is_source_of relationship

The District instance will contain the specific implementation plan, not just a file reference."

Important Notes

• Source files stored in instance folders
• Content extracted and saved as .md files
• Relationships managed in root graph_edges
• Metamodel defined only in root file
• Enrollment explicit via .iNNfo.md files