CortexMark Pipeline — VS Code Extension

A VS Code extension for running the CortexMark PDF → Markdown pipeline from inside the editor under the marketplace display name CortexMark Pipeline.

It provides:

• session-based PDF processing,
• pipeline and analysis commands,
• Markdown preview,
• dashboard metrics,
• real-time progress and logging,
• a chat-oriented control surface.

Important: this extension is the UI layer only. It does not bundle the Python backend. Install cortexmark separately.

Install

Install from the marketplace

1. Open VS Code
2. Open the Extensions view
3. Search for CortexMark Pipeline
4. Install the published extension ID: PhiniteLab.cortexmark-pipeline-vscode

Install from VSIX

You can also install a .vsix package via:

Ctrl/Cmd + Shift + P → Extensions: Install from VSIX...

Backend requirements

The extension launches the Python backend in your workspace environment.

Minimum backend

pip install cortexmark

Recommended backend for academic PDFs

pip install torch torchvision --index-url https://download.pytorch.org/whl/cpu
pip install "cortexmark[docling]"

Optional system tools

• poppler-utils / poppler
• tesseract-ocr / tesseract

These are useful for OCR-heavy or image-heavy PDF workflows, but they are not required for every project.

Requirements at a glance

What you can process

Primary user input inside the extension:

• one or more PDF files
• a folder containing PDFs

Outputs you can inspect after processing:

• raw Markdown
• cleaned Markdown
• chunk files
• semantic chunk files
• quality reports
• previewable Markdown pages
• dashboard summaries

First-time setup

1. Open your workspace folder in VS Code
2. Install the extension
3. Install the Python backend in the interpreter you want the extension to use
4. Run CortexMark: Environment Doctor
5. If needed, run CortexMark: Setup Wizard
6. Create a session
7. Add PDFs or add a PDF folder
8. Run CortexMark: Run Full Pipeline

Daily workflow

1. Create or select a session

Sessions keep inputs and outputs isolated. The extension stores session metadata in:

.cortexmark/sessions.json

Session-scoped pipeline artifacts live under:

sessions/<session-name>/
├── data/raw/
└── outputs/

2. Add PDFs

Use either:

• Add PDFs...
• Add PDF Folder...

3. Run commands

Useful commands include:

• Run Full Pipeline
• Convert Only
• Generate QA Report
• Run Cross-Reference Analysis
• Run Algorithm Extraction
• Run Notation Glossary
• Run Semantic Chunking
• Preview Markdown
• Refresh Dashboard

Settings

| Setting | Default | Description | |---|---|---| | cortexmark.pythonPath | python3 | Python executable override | | cortexmark.configPath | configs/pipeline.yaml | Pipeline config path | | cortexmark.dataRoot | | Optional input root override | | `cortexmark.outputRoot` | | Optional shared output root override | | cortexmark.sessionStorePath | .cortexmark/sessions.json | Session metadata path | | cortexmark.defaultEngine | dual | Default conversion engine | | cortexmark.autoProcess | false | Auto-process new PDFs |

Path resolution precedence:

1. explicit VS Code setting (cortexmark.*)
2. process environment variables
3. workspace .env
4. selected pipeline config
5. workspace-relative defaults

Commands

All commands are available from the Command Palette with the CortexMark: prefix.

Key commands:

• Environment Doctor
• Setup Wizard
• New Session
• Process Active Session
• Run Full Pipeline
• Convert Only
• Generate QA Report
• Run Cross-Reference Analysis
• Run Algorithm Extraction
• Run Notation Glossary
• Run Semantic Chunking
• Preview Markdown
• Refresh Dashboard

See also:

• https://github.com/PhiniteLab/pdf-to-markdown-pipeline/blob/main/docs/vscode/setup.md
• https://github.com/PhiniteLab/pdf-to-markdown-pipeline/blob/main/docs/vscode/commands.md

Development

cd vscode-extension
npm install
npm run compile

Press F5 in VS Code to launch an Extension Development Host.