BlackBoxApproach: Building Apps When AI Keeps Breaking Things
Proof-of-concept research. github.com/amiune/theblackboxapproach
AI is great at generating code. It's also great at rewriting your entire project when you asked it to tweak one button. BlackBoxApproach is our experiment in fixing that.
What it tries to solve
Human vs AI decisions. Every box and data field tracks provenance: human_defined, ai_defined, or mixed. Once you confirm something, AI can't downgrade it back to a guess.
Pseudocode as a middle layer. Boxes use hybrid pseudocode (CALL, IF, natural language): enough structure to decompose a system, not enough to drown in implementation details until you're ready.
Modular boundaries. Refine one black box at a time. Ancestors are read-only context; only the selected subtree can change. The visualizer generates prompts that enforce this, so "fix the payment flow" doesn't rewrite your auth layer.
A human-readable view. A native macOS visualizer shows the box tree, provenance badges, and contracts. Simpler than dozens of scattered spec files.
How it works
Each app project gets a blackboxes/ folder: YAML boxes, shared data structures, a skill file the LLM follows when explicitly invoked, and a provenance log. ChatGPT (or similar) returns structured file updates; the app writes them, tracks git diff, and blocks new AI edits while you have uncommitted changes.
Not low-code. Not "natural language programming." Recursive black-box decomposition with contracts, intent, and audit trail.
Status
Early POC. macOS app + project template. The method matters more than the UI. The goal is disciplined AI-assisted development that stays understandable to humans.