# TOOLS.md — Tatiana

## What I Use

### Image Generation
- **Gemini API** — Primary generation tool
  - Env: `GEMINI_API_KEY` (set in `~/.bashrc`)
  - Models: `gemini-3.1-flash-image-preview` (default), `imagen-4.0-generate-001`, `imagen-4.0-ultra-generate-001`
  - Skill: Nano Banana Pro (`skills/` — check REGISTRY.md)
- Generation scripts in `scripts/images/`:
  - `gen-*.py` — Various image generation scripts (adapt per project)
  - `image_processor.py` — Processing pipeline
  - `image-postprocess.py` — Post-generation adjustments
  - `composite-panel.py` — Multi-image panel compositing
  - `composite-slide.py` — Slide image compositing

### Screenshots & References
- **Playwright** (`/usr/bin/playwright`) — Capture reference pages, verify visual output
- `scripts/taste/taste-screenshots.py` — Taste board reference captures
- `scripts/images/image_analyzer.py` — Analyze reference images

### Image Processing
- `scripts/images/compress-images.js` — **Always compress before delivery** (`node scripts/images/compress-images.js <dir>`)
- `scripts/images/upload-image.cjs` — Upload to hosting
- `scripts/images/upload_slide_image.py` — Upload for slide use
- `scripts/images/crop-github-header.py` — Crop utility

### Standards
- `agents/anton/VISUAL-STANDARDS.md` — Visual quality bar for CE work
- `agents/anton/STANDARDS.md` — CE brand system (colors, typography)

### Taste Board (Reference Only)
- Query taste board for style references before starting visual work
- `scripts/taste/extract-taste-dna.py` — Understand visual DNA of references

## What I Don't Use
- No copywriting (that's Ogilvy)
- No layout systems (that's Jessica)
- No research/scraping (that's Julia)
- No deployments (that's Thibault)