WEIPING_LAB

An evidence-gated research workbench for discovery, falsification, experiment handoff, and audit.

Purpose

WEIPING_LAB is an autonomous research workbench for phenomenon-driven discovery, kill-first ideation, experiment planning, evidence-gated writing, and cross-model audit. Its central design goal is to preserve the boundary between a plausible research plan and evidence produced by an actual experiment.

Workflow

The repository organizes a gated path from phenomenon or project context through novelty review, adversarial falsification, experiment planning, a human/server experiment bridge, results loading, writing, citation review, and claim audit. Runtime status is observable through its CLI and API, while provider credentials and research workspace state remain local.

Project banner from the public WEIPING_LAB repository.

Evidence and boundaries

WEIPING_LAB is workflow software, not evidence that an experiment has run or a paper claim is true. The repository labels results as paper_result, official, diagnostic, or pilot; only evidence that passes the configured gates is eligible for main claims. The workflow deliberately pauses where real experiments or human decisions are required.

Artifacts and provenance

  • Source code and documentation
  • Displayed banner source
  • The project is released under the repository’s MIT license. Open-source systems studied during its design remain credited in the repository documentation; they are not vendored or claimed as original work here.