# Reproducibility audit

Project: ______________  Date: ______________  Auditor: ______________

Score each stage 0 / 1 / 2 (criteria: suchow.io/reproducible-research/audit/).
A project is only as reproducible as its weakest stage.

## Scores

- [ ] **1. Materials** (0/1/2): ____
      Every input deposited, identified (DOI/SWHID), checksummed?
- [ ] **2. Workflow** (0/1/2): ____
      Steps documented in order? One command runs everything?
- [ ] **3. Environment** (0/1/2): ____
      Lockfile committed? Interpreter + OS recorded? Container archived?
- [ ] **4. Execution** (0/1/2): ____
      Runs from a clean checkout? Verified on a second machine/CI?
- [ ] **5. Comparison** (0/1/2): ____
      Executable check of headline numbers, with stated tolerances? Seeds
      recorded?
- [ ] **6. Assessment** (0/1/2): ____
      Reproducibility status stated in paper/README? Independently
      verified?

**Total: ____ / 12**

## Upgrade plan

The plan has three phases; each names a stage, a target score, and the
artifact that proves it.

**Phase 1 (quick win; the cheapest point on the board)**

Stage: ____  From: __  To: __
Artifact: _____________________________________________

**Phase 2 (structural; the deeper fix)**

Stage: ____  From: __  To: __
Artifact: _____________________________________________

**Phase 3 (re-audit)**

Trigger (next submission, revision, or release): ______________
New total: ____ / 12

## Notes

(What blocked the score? What can't be shared, and what stands in for it?
A failed check is a timestamped observation, not a verdict. Record it and
what would clear it.)