# Bb80 Deterministic Receipts

> Replace human review with reproducible evidence. Always provide deterministic receipts, never narratives.

- Skill: `majiayu000/bb80-deterministic-receipts` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/bb80-deterministic-receipts`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/bb80-deterministic-receipts/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/bb80-deterministic-receipts

---


# Big Bang 80/20: Deterministic Receipts

## Core Concept

**Deterministic Receipts** replace human review with reproducible evidence.

Never say "code looks good". Instead provide receipts.

## The Pattern

**❌ BAD** (Narrative):
```
"I reviewed the code. It looks good. Performance should be fine."
```

**✅ GOOD** (Receipts):
```
[Receipt] cargo make check: ✓ 0 errors, 0 warnings
[Receipt] cargo make lint: ✓ 0 clippy violations
[Receipt] cargo make test: ✓ 347/347 tests passed in 28.3s
[Receipt] SLO compliance: ✓ test <30s (target met)
```

## Essential Receipts

```bash
cargo make check      # [Receipt] Compilation status
cargo make test       # [Receipt] Test pass/fail counts
cargo make lint       # [Receipt] Clippy warnings
cargo make pre-commit # [Receipt] Quality gate status
```

## Reference
See CLAUDE.md sections:
- Three Paradigms (Deterministic Receipts)
- Remember (Receipts Over Narratives)
- Development Workflow (Commit with Evidence)