Good AI Bad AI
Smoketest is live
This baseline confirms we can ship quickly and iterate into the first playable puzzle level.
Checks
- SvelteKit app booted
- Cloudflare Worker adapter configured
- Wrangler deployment pipeline ready
- AI context/style/level scaffolding added
Next target: Level 1 (Navigation)
- Realtime reward-button feedback loop
- Single-agent navigation from center to exit
- Simple score with local storage persistence
See ai/levels/level-01-navigation.md and ai/prompts/ for scaffolding details.
Playable proof of concept
playable Dodging is now available as an early gameplay loop.
Open /levels/dodging to play the first dodging prototype.
Scaffolded styles
- silly-friction: Readable objectives with awkward, funny execution.
- toy-lab: Simple training-lab framing with visible state and feedback.
Style source of truth: ai/styles/style-profiles.yaml
Level manifest
- shipped Smoketest — Validate runtime and deployment path.
- scaffolded Navigation — Train one agent from center to right-side escape.
- playable Dodging — Proof-of-concept loop with hazards, reward/punish, and reset.
Current build target: Navigation (ai/levels/level-01-navigation.md)