Help & reference
Answers to common questions about how PASS works and how to get the most out of an analysis, plus where to reach the team.
Overview
PASS takes a manuscript (JATS XML, PDF, or pasted text) and produces a ranked list of candidate journals along with the evidence behind each pick. The full method is laid out on the Method page.
Getting started
- Sign in. Use Google or create an account from the login page.
- Add your OpenAI key (optional). On the Settings page. Your own key gives unlimited runs; without one, PASS uses the shared key with a daily limit. Keys are encrypted at rest.
- Submit a manuscript. On the home page, drop a PDF or bioRxiv/medRxiv JATS XML into the box, or paste the title and abstract. Choose the candidate journals next to it and press Run.
- Read the report. Analyses finish in about 15 minutes. You'll get an email when yours is ready.
Frequently asked questions
What file formats are supported?
JATS XML (the bioRxiv/medRxiv preprint format) gives the cleanest parsing. PDFs also work, where the parser extracts title, abstract, and main sections. You can also paste the title and abstract directly if you'd rather skip uploading a file.
How does the system pick journals?
The pipeline is composed of three agents: a manuscript-understanding agent that builds a structured representation of the paper, a literature-retrieval agent that assesses overlap and positioning relative to prior work, and a journal-fit reasoning agent that integrates both to produce a ranked list of candidate journals. The paper is scored across novelty, writing quality, conceptual soundness, and impact potential, then matched to journals using both quality and topic fit to the inferred subject areas.
What does the date cutoff mean?
PASS uses this date to classify retrieved papers as prior art (published before), concurrent work (within the grace window after), or future context (after the grace window). It's auto-extracted from your uploaded manuscript. To override it, for example, if you submitted your paper a while ago and want recommendations based on what the field looked like at that time, open the settings gear next to Run on the home page and set the Submission date. It defaults to today.
Why do different runs sometimes give different rankings?
The LLM is not deterministic. The same input can yield slightly different scores or wording on different runs. To reduce noise, every analysis runs the final ranking step three times and the report is ensembled across them.
How long does an analysis take?
Typically about 15 minutes end-to-end. You can close the tab and come back — you'll get an email when it's done.
Who can see my uploaded manuscripts?
Only you (and the platform admins). Manuscripts are stored in your account.
Can I re-run an analysis?
Yes. Just submit the same manuscript again from the New analysis page. Each submission gets a fresh paper id so you can compare runs.
I have a specific candidate journal list, how do I build it?
Open the Candidate journals box on the home page. It offers prebuilt lists from SCImago: a main subject area, on its own or together with Multidisciplinary, cut to the top N journals by SJR (default 100). SJR (SCImago Journal Rank) is a citation-weighted prestige metric, so top-N keeps the N most-prestigious journals in those areas. Hover a list to preview it. To build your own, open My candidate journal list and click New list: start from a SCImago top list, an empty list, pasted names, or a CSV, then add journals by name or remove any, and save. Saved lists appear under Your lists in the same box. The agent only recommends journals from the list you pick; choose No shortlist to let it recommend from the whole field.
How do I delete an analysis?
From your dashboard ("My papers"), each analysis has a small delete control. This hides the analysis from your view but keeps the row + any feedback you submitted in the database for research purposes. Contact the team if you need a hard delete.
Contact & Support
For questions, feedback, or technical support, please contact us:
- The paper id of the analysis (visible in the URL)
- A short description of what you expected vs what happened
- Screenshots if applicable
Citation
If PASS helps with a paper you publish, we'd love a citation. Please cite the preprint:
Chen J, Zhang Z, Li B, Sun Q, Zhang Y, Tam E, Lin J, Li D, Li Y, Zhao B. Positioning manuscripts in the scientific landscape with agentic AI. arXiv:2609.13760 (2026). https://arxiv.org/abs/2609.13760
@misc{chen2026positioningmanuscriptsscientificlandscape,
title={Positioning manuscripts in the scientific landscape with agentic AI},
author={Jiawen Chen and Zichen Zhang and Bingxuan Li and Quan Sun and Yiyan Zhang and Edric Tam and Jinjie Lin and Didong Li and Yun Li and Bingxin Zhao},
year={2026},
eprint={2609.13760},
archivePrefix={arXiv},
primaryClass={cs.AI},
url={https://arxiv.org/abs/2609.13760},
}
Feel free to reach out for any additional information.