paper-workbench
Run the following command in DeepSeek Harness:
dsh plugin install LessXi/paper-workbench
Paste the following prompt into your AI chat to install this plugin:
In a DeepSeek Harness terminal run dsh plugin install LessXi/paper-workbench. The plugin source is https://github.com/LessXi/paper-workbench. Restart the session after installation to use /paper-init and other commands.
About this plugin
The hardest part of reading a paper is not finding it but finishing a forty-page PDF and retaining nothing. paper-workbench is built against that gap: its learn mode refuses to read for you. The AI sketches a study map, answers questions, asks you to recap without notes and then corrects mistakes, writes quiz questions, and reviews your code. Equations and code must pass through your own thinking.
The workflow spans the full arc from literature search to reproduction: multi-source retrieval with must-read, optional, and skip tiers; MinerU-based close reading with LaTeX formula restoration and table extraction; three-pass structured notes; citation graphs; method family trees with evolution timelines; arXiv watch digests; and CPU-level reproduction from down-scaled official repo through from-scratch implementation to numerical cross-check. Parsing offers a lite tier ready in seconds and a full tier for precise reading, upgradeable anytime.
Aimed at graduate students and researchers who want an AI sparring partner rather than a ghostwriter. Works across ZCode, DeepSeek Harness, and Claude Code agent environments. The workspace doubles as an Obsidian vault, so Sync covers everything with zero extra configuration.
Use Cases
- Close-read a paper where AI sketches a study map, quizzes you, and reviews your code instead of writing summaries
- Build a method family tree, comparison matrix, and evolution timeline before you start a survey
- Reproduce experiments on CPU: down-scaled official repo, from-scratch implementation, numerical cross-check
Best For
- Graduate students who read papers and write surveys and want retention, not summaries
- Researchers reproducing paper experiments on local CPU hardware
- Self-directed learners who treat AI as a sparring partner, not a ghostwriter
Related Plugins
A service-oriented sidebar framework for DSH, offering a full workbench with file explorer, embedded browser, real terminal, Git panel, and extensible plugin services.
A beautiful, practical Claude Code-style TUI plugin with pixel whale top bar, flowing glow title, real-time status line, streaming thought expansion, time rewind, context progress bar and TPS gauge — zero core changes.
The plugin market for DeepSeek Harness: browse/search/one-click install community plugins and themes, with updates, backup/restore, hot disable, diagnostics, and AI fix.
A persistent whale widget on the DSH web UI that shows DeepSeek balance, today's usage, and per-turn cost, with drag-and-snap and sound effects.