Literature Paper Breaker

Literature Paper Breaker is an AI-assisted research and paper-writing workspace for humanities and social science projects.

This repository is a cleaned-up, source-first reconstruction of the original LPB demo bundle. It keeps the Python source, web UI, review workflow, and local methodology skills, while removing bundled wheel caches and generated demo outputs from version control.

What It Does

• runs multi-stage research jobs from a topic to a draft paper
• builds literature reviews from multiple academic sources
• generates outlines, proposals, and manuscript drafts
• supports peer review, revision tasks, annotations, and author revision rounds
• provides a browser-based editorial workroom for reading, reviewing, and revising papers
• can generate academic-style figures for supported papers

Repository Layout

• src/: application source code
• app_skills/: local methodology and editorial skills used by the site
• scripts/: local startup helpers
• tests/: regression tests for the reconstructed source repo
• papers/: runtime output directory for generated paper workspaces
• docs/RECONSTRUCTION_NOTES.md: notes about what was reconstructed and what is still missing
• docs/PRODUCT_ARCHITECTURE.md: product-facing architecture diagram and explanation

Product Architecture

For a user-friendly system overview, see docs/PRODUCT_ARCHITECTURE.md.

Product Visuals

User Journey

Peer Review Workflow

Quick Start

Create a Python 3.10 environment and install the project in editable mode:

py -3.10 -m venv .venv
.\.venv\Scripts\python.exe -m pip install --upgrade pip
.\.venv\Scripts\python.exe -m pip install -e ".[llm]"

Then start the local site:

.\.venv\Scripts\python.exe -m src.cli site -o papers --port 8092

Or use the bundled helper:

powershell -ExecutionPolicy Bypass -File .\scripts\start_source.ps1 -Port 8092

Open http://127.0.0.1:8092.

Environment

Copy .env.example to .env and set the LLM provider keys you want to use.

The current reconstruction works well with OpenRouter-backed OpenAI-compatible models such as gpt-5.4.

Running Tests

$env:PYTEST_DISABLE_PLUGIN_AUTOLOAD='1'
.\.venv\Scripts\python.exe -m pytest tests -q

Current Status

This is not the original upstream development repository.

What is included:

• core source code
• web UI and CLI
• workroom, review, and skills flows
• reconstructed docs and regression tests

What is still missing:

• original git history
• complete upstream methodology source library
• original CI/CD setup
• some original packaging and authoring context

Output Policy

Generated paper workspaces under papers/ are treated as local runtime output and are intentionally not tracked in git.

License

MIT. See LICENSE.