Priorwork

Hand your literature review to an AI agent, then check it — no terminal needed.

日本語

Priorwork keeps a topic-based literature review in a Git workspace. Tell your AI agent (Claude Code, Antigravity or any agent that reads AGENTS.md) the topic, in plain words: it sets the scope, searches Semantic Scholar and OpenAlex, screens with a reason for every decision, chases citations, fills in a card for each paper, writes the review, checks and exports it — without stopping. This extension is where you check and correct its work, without a terminal. The extension never calls an AI service itself.

• Ask your agent. New Survey → Ask your agent to do it copies a request for a whole survey; Ask Your Agent… on a survey copies one to carry on, search more, screen, chase citations, fill in the cards, write or check. A button opens the chat of Claude Code, Antigravity or Copilot when it is installed. You can also just ask in the chat — no skill names needed.
• Surveys in the sidebar. Each survey with what is left to screen, its next steps, and its papers grouped as to screen, maybe, included and excluded.
• A page per paper. Title, authors, journal and its SSCI status, citations, warnings (a DOI that points to another paper, a working-paper version, a possible book review) and the abstract, with the survey's inclusion and exclusion criteria at the top and the reason the agent gave. Overturn a decision with a click or a key — I include, M maybe, X exclude (with a reason). For included papers, the page also holds the paper card (research question, X, Y, data, identification, findings, limitations and how it was checked): check and correct what your agent wrote, and save with Ctrl+S.
• Search and snowball. Search in English (by relevance, boolean bulk search, or SSCI journals only), then chase the references and citations of the included papers to find what keyword search missed.
• Checks you can trust. Check the Survey lists author–year citations in the text that match no registered paper, DOIs that point to other titles and unfilled cards in the Problems panel.
• Reports inside VS Code. Export and View the Report, or right-click reports/*.html → View the Report in Priorwork, shows the report as it will be read (marked Draft until it is done) — on a server over SSH too. When the agent exports it, a notification offers to show it.
• Zotero, read only. Link a Zotero collection to a survey: included papers missing from it are one Copy the DOIs away from Zotero's magic wand, papers you put in it can be registered as candidates, and full texts come from its PDFs first.
• Settings without files. API keys, Zotero and the SSCI list on one page, with a connection check; keys are never shown. A row in the sidebar publishes the workspace to a private GitHub repository.

Getting started

1. Install the extension. Nothing else is needed to start: it uses Python 3.10 or later if you have it, and otherwise installs uv, which downloads Python (into your home folder, without administrator rights). You will also want an AI agent that works in VS Code (Claude Code, Antigravity, …), and git to keep the workspace on GitHub.
2. Run Priorwork: Create a Workspace… (or use the button in the Priorwork sidebar). Pick an empty folder and the language of the reports (English or Japanese). The extension creates a .venv there and installs the priorwork command from PyPI.
3. Open the Settings (the gear in the sidebar) and add a free Semantic Scholar API key (searches without one are often rate-limited). Zotero and the SSCI list are optional and set there too; Save and Check the Connections tells you whether they work.
4. New Survey… → Ask your agent to do it: give the topic and the depth, and paste the copied request into your agent's chat (or just ask in the chat).
5. Read the report when the agent has exported it, then check the decisions and cards on each paper's page and correct what you disagree with. The agent follows your changes.
6. Ask for more with Ask Your Agent… on the survey. To search and screen yourself instead, choose Set it up myself in step 4 and use Search… and Screen Candidates.

Keep the workspace in a private GitHub repository (the row Not on GitHub yet in the sidebar sets it up): the reports and state files contain abstracts. .env, which holds your keys, is kept out of Git.

Commands

Settings

What leaves your machine

Search queries and DOIs go to Semantic Scholar and OpenAlex; with Zotero set up, the extension reads your Zotero library (read only). Nothing else is sent anywhere.

More

Everything — the steps of a survey, the paper cards, the settings, the agent, the command-line tool — is in the guide; a shorter overview is the project's README. The extension and the priorwork command are released together, under one version number (changelog).