reproproof

Product decision

Decision date: 2026-07-11. Scores are comparative judgments from 1 (weak) to 5 (strong), not measured market facts.

Candidates

Criterion ReproProof: issue-to-reproduction verifier Dependency-upgrade evidence bot Release-notes and migration verifier
Problem severity 5 4 3
Plausible user breadth 4 5 3
Differentiation 4 2 3
Solo maintainability 3 4 4
Contributor-friendly modules 5 4 3
Downstream dependency potential 4 3 2
Security risk (5 = lower) 2 3 4
MVP feasibility 3 5 4
Reason for repeated use 5 4 3
Total 35 34 29

Decision

Build ReproProof, an evidence-first issue-to-reproduction CLI and GitHub Action.

The margin over dependency upgrades is deliberately small. Upgrade automation is easier to ship, but Dependabot, Renovate, and platform-native tooling already cover much of that workflow. Issue reproduction is harder and riskier, yet the maintained research and tooling landscape shows an unresolved gap between free-form bug reports and verified failing tests.

Positioning

ReproProof does not promise autonomous bug fixing. It produces:

  1. a structured reading of expected/actual behavior, steps, environment, missing details, and assumptions;
  2. a reviewable candidate test patch;
  3. execution evidence from an isolated copy;
  4. an explicit status and confidence level; and
  5. Markdown and JSON artifacts usable by humans and later automation.

Differentiation

MVP boundary

The v0.1 line supports local Markdown or opt-in public GitHub issue URLs, local repositories, Node and Python adapters, node:test/Jest/Vitest/pytest command selection, mock/OpenAI/Anthropic/loopback providers, locked-down Docker execution, an explicitly unsafe local fixture mode, reports/patches, and double-opt-in Draft PR creation with no merge path. Docker and GitHub writes are implemented but await live infrastructure validation; production-grade benchmark performance remains future work.

Falsification criteria

Reconsider or narrow the project if, after ten substantive maintainer interviews and three opt-in pilots, maintainers consistently report that setup/review costs exceed saved triage time, or if wrong-reason reproductions cannot be kept acceptably low with transparent evidence.