| name | interview-visual-assets |
| description | Plan, add, and generate visual assets for behavioral interview specs in this repository. Use when deciding where interviews should be illustrated, adding assets.icon, explainerComic, strong/weak answer pattern aiVisual, strong/weak signal visuals, or exampleStories aiVisual fields, running dry-run image generation scripts, or improving visual prompts and generated interview imagery. |
Interview Visual Assets
Workflow
- Read the target
interview.json and inspect whether it already has assets.icon, explainerComic, strongAnswerPattern.aiVisual, weakAnswerPattern.aiVisual, evaluation.signals[].strongAiVisual, evaluation.signals[].weakAiVisual, or exampleStories[].aiVisual.
- Decide the useful illustration surfaces before generating files:
- Overview icon for quick scanning.
- Explainer comic for the interview's central tension.
- Strong/weak answer-pattern visuals for the answer contrast.
- Strong/weak signal visuals for evaluation calibration.
- Example-story visual for concrete, memorable variants.
- Prefer visuals that clarify behavior, trade-offs, or operating mechanisms. Avoid decorative images that only restate the title.
- Use dry-run mode first and review planned paths/prompts.
- Generate assets only when the user wants real images and
GEMINI_API_KEY is available.
- Rebuild the generated site after JSON or asset changes.
Image Fields
Use dataset-relative paths:
assets.icon: small overview image for the card.
explainerComic: visual summary image for the interview.
strongAnswerPattern.aiVisual: visual attached to the strong answer pattern.
weakAnswerPattern.aiVisual: visual attached to the weak answer pattern.
evaluation.signals[].strongAiVisual: visual attached to the strong side of an evaluation signal.
evaluation.signals[].weakAiVisual: visual attached to the weak side of an evaluation signal.
exampleStories[].aiVisual: visual attached to a concrete example story.
The static templates make generated images clickable and open them in a new tab.
Scripts
Use these project scripts from the repository root:
python3 _scripts/generate_missing_interview_icons.py --dry-run
python3 _scripts/generate_interview_comic.py --dry-run
python3 _scripts/generate_answer_pattern_pictures.py --dry-run
python3 _scripts/generate_signal_pictures.py --dry-run
python3 _scripts/generate_example_story_pictures.py --dry-run
python3 _scripts/generate_interview_assets.py --dry-run
For real generation, rerun the relevant script without --dry-run only after confirming credentials and desired scope.
Verification
Run:
python3 _scripts/validate_interviews.py
python3 build.py
python3 _scripts/smoke_static_site.py