Visual daily work
Question Bank, Editor, Paper, Import, Data, and Quality workspaces with offline KaTeX and structured tables.
The complete life of a question
Questions move through explicit, reviewable stages. Source evidence and human decisions remain visible instead of disappearing behind one opaque import action.
- 01ImportAdapt JSON, Markdown, LaTeX, DOCX, or a source-specific pipeline.
- 02StructureNormalize text, formulas, tables, choices, answers, and metadata.
- 03ReviewInspect candidates and preserve an explicit human acceptance boundary.
- 04Edit and checkWork field by field, then run deterministic local quality checks.
- 05Assemble and exportBuild a paper and export canonical JSON or conventional DOCX.
Product workspace, stable core
The downloadable application and the Python contracts share one canonical model. Users can start visually while contributors extend formats, storage, and workflows.
Replaceable boundaries
Canonical schema, format registry, plugin discovery, storage adapters, compatibility fixtures, and stable public APIs.
Review before persistence
AI providers and future document pipelines produce reviewable candidates; credentials and silent production writes are not bundled.
Visual quick start
Open the public workspace with one command.
Python 3.10 or newer is the only requirement. The server binds to loopback, creates an ignored local workspace, and opens the reviewed synthetic case.
The repository ships no production question bank, private correction rules, credentials, provider configuration, or operational records.
git clone https://github.com/wzsisshadiao-crypto/dq-questionbank-core.git
cd dq-questionbank-core
python run.py