seanfzc/openclaw-workspace

Workspace for the ATOM-SG project — a Recognition-First Integrated Training System for P6 Mathematics in Singapore.

Summary

Contains exam paper processing pipelines, math problem banks, a FastAPI backend, HTML5 frontend, OCR/rendering systems, and Obsidian dashboards for tracking progress. Two implementation tracks exist: ATOM-SG Pilot/ (MVP v4.1) and ATOM_SG_v6_2/ (dual-pass pipeline).

Included Assets

Documents & Configs

• AGENTS.md — Agent session startup and memory guidelines
• CLAUDE.md — Behavioral coding guidelines

ATOM-SG Pilot (MVP)

• 00-Original-Exam-Papers/ — 5 P6 prelim PDFs (ACS Junior, Henry Park, Nan Hua, Nanyang, Raffles Girls)
• 03-WordProblems/ — 20 word problems across 4 pathways (Before-After, Part-Whole, Cross-Thread, Data Interpretation)
• 04-OCR/ — Tesseract OCR pipeline for diagram text extraction
• 05-Backend/ — FastAPI backend with requirements.txt
  • frontend/ — HTML5/JS SPA (no build tools)
  • rendering/ — Grid and isometric geometry diagram renderers (Matplotlib)
  • playwright-tests/ — Playwright E2E test suite with npm package.json
• 06-Dashboards/ — Obsidian Dataview templates for milestones, risks, problem bank
• 07-Backups/ — Git-backed backup automation
• deployment/ — Docker Compose deployment

ATOM_SG_v6_2

• SOUL.md / AGENTS.md / TOOLS.md — Agent identity and pipeline definitions
• DUAL-PASS "NEO" PIPELINE — Document processing using opendataloader + pymupdf4llm + DeepSeek formula conversion

How to Use

Backend

cd ATOM-SG\ Pilot/05-Backend
pip install -r requirements.txt
python main.py
# Available at http://localhost:5000

Frontend

Served automatically from backend at /static/*. Visit /docs for Swagger UI.

Playwright Tests

cd ATOM-SG\ Pilot/05-Backend/playwright-tests
npm install
npm run install:browsers
npm run test:critical  # Run critical tests first

Rendering Diagrams

from rendering.grid_renderer import GridRenderer
renderer = GridRenderer(output_dir="./outputs")
renderer.process("spec.yaml", is_file=True)

OCR Pipeline

python3 ocr_pipeline.py /path/to/problem.pdf --output-dir ../artifacts/ocr --prefix Problem-001

Docker Deployment

cd ATOM-SG\ Pilot/deployment
docker-compose up -d

Notes

• Source of truth for exam questions: 00-Original-Exam-Papers/ (do not use deprecated exam_samples/)
• Python 3.x required; Tesseract OCR required for diagram extraction
• Frontend is vanilla HTML/JS — no build step needed
• Obsidian dashboards require the Dataview plugin