swangek/Data-Visualization-Pipeline

Full sync: two-phase vis code (Phase 1 skill-only + Phase 2 evolution), modular architecture, stat vis cookbook

Two-phase vis code: Phase 1 (skill-only general code) + Phase 2 (evolve with real data)

Rewrite mixed+evolution prompts: direct data-specific code gen instead of generic processor

Improve evolution + mixed prompts: require generic logic, no hardcoded column checks

Treat 0 charts output as execution failure to trigger evolution

Improve mixed prompt: semantic-aware visualization with statistical chart guidance

Update frontend: evolution badge + log display

Refactor: modular architecture + universal vis + code evolution

Auto-generate vis code if missing + auto-regenerate if execution fails

Fix: vis code auto-detects JSON key + longer LLM health timeout

Fix: vis code auto-detects JSON key + longer LLM health timeout

Add LLM health check indicator

Add LLM health check indicator

Fix KeyError: use .get() for chart fields from LLM-normalized specs

Remove hand-written skill spec — should be generated by LLM via upload-skill

Fix quantitative_analysis skill: use std/min/max/median instead of sum/value

Add python-multipart to requirements.txt

Add quantitative_analysis skill + sample data

Add quantitative_analysis skill + sample data

Add quantitative_analysis skill + sample data

Fix README emoji encoding

Fix Dockerfile: add user permissions for HF Space

Deploy Data Visualization Pipeline with Grok API + Plotly HTML output

initial commit