System Architecture

Expert Council (7 Agents)

This project implements a Council of 7 Specialized Experts that collaboratively develop, validate, and deploy scientific content:

flowchart LR
    A["@Architect\nDesign"] --> B["@Scientist\nTheory"]
    B --> C["@Engineer\nCode"]
    C --> D["@Safety_Gate\nValidation"]
    D -->|"✓ Safe"| E["@Analyst\nAnalysis"]
    D -->|"✗ Error"| C
    E --> F["@Librarian\nExperimental Validation"]
    F --> G["@QA\nAudit"]
    G -->|"✓ Pass"| H["Publish"]
    G -->|"✗ Fail"| B

Agent Roles

Agent	Responsibility	External Skills
@Architect	Structure guardian and project memory	senior-architect, agent-memory-systems
@Scientist	Theory owner, LaTeX notation	claude-scientific-skills, research-engineer
@Engineer	Code builder, implementation	python-pro, ml-pipeline-workflow
@Safety_Gate	Numerical validation, toxicology, pedagogy	stability_guardian, toxicity_predictor, socratic_debugger
@Analyst	Deep analysis and visualization	data-storytelling, descriptor_miner
@Librarian	Experimental validation (Materials Project)	librarian_rag
@QA	Supreme quality auditor	systematic-debugging, code-review-excellence

External Skills

Modular skills developed for scientific validation:

Numerical Skills

• stability_guardian.py — MD timestep validator
• basis_set_architect.py — Gaussian basis set recommender for DFT

AI Mining Skills

• toxicity_predictor.py — Molecular toxicity predictor

Pedagogy Skills

• socratic_debugger.py — Socratic pedagogical feedback generator

Orchestration Skills

• librarian_rag.py — RAG for experimental validation

Why Python 3.11?

Library	Python 3.10	Python 3.11	Python 3.12
RDKit	✓ Stable	✓✓ Optimal	⚠️ Beta
ASE	✓	✓✓	✓
OpenMM	✓	✓✓	❌

Python 3.11 offers maximum compatibility with the complete scientific stack used in this course.