Your library, in your terminal.
Full-text papers, preprints and trial records for your own agents and tools. Add the MCP server and ask from Claude Code, or call the API from your lab’s code.
$ claude mcp add --transport http valyu https://mcp.valyu.ai/mcp ✓ Added MCP server valyu > How does exercise compare with SSRIs for depression? A network meta-analysis of 218 RCTs found walking, yoga and strength training gave moderate effects, comparable to psychotherapy and SSRIs. [1] [1] Noetel M et al. BMJ, 2024.
Systematic reviews
A search log your reviewers can reproduce.
Every Valyu search is logged with its query, source, date and the results it returned. That is the identification stage of your PRISMA 2020 flow diagram, documented as you go.
Screening and eligibility stay in the tool your team already uses, such as Covidence or Rayyan, or in an agent you build on the API.
| Date | Source | Query | Results |
|---|---|---|---|
| 2026-09-22 | PubMed | exercise AND depression AND randomised controlled trial | 20 |
| 2026-09-22 | PubMed | physical activity intervention major depressive disorder randomized trial | 20 |
| 2026-09-22 | PubMed | yoga OR resistance training depressive symptoms RCT | 20 |
60 records from three PubMed searches, run through the Valyu API on 22 September 2026. Each search returned its top 20 results, and the query and date for every search are in the log above.
Writing
Cite the paper, not the tool.
Every result carries its full bibliographic record, so it goes straight into your reference manager.
@article{noetel2024exercise,
title = {Effect of exercise for depression: systematic
review and network meta-analysis of
randomised controlled trials},
author = {Noetel, Michael and others},
journal = {BMJ},
year = {2024},
volume = {384},
pages = {e075847}
}Lab agents & alerts
An agent that has read your field.
Give your lab’s agents full-text access, or rerun a query each week and get only what is new.
from valyu import Valyu
valyu = Valyu()
res = valyu.search(
"long-read sequencing structural variants",
start_date="2026-09-15", # only what is new
)Proof
Benchmarked on literature synthesis.
ScholarQA was developed by researchers at the University of Washington, Allen Institute for AI, Meta and Carnegie Mellon.
ScholarQA-CS
Literature synthesis on 110 computer-science questions, graded against expert rubrics. Published in Nature, 2026.
Pricing
Start free.
$10 in free credits when you sign up, or $20 with a work email, usable across every source. After that, pay as you go from $10.
FAQ
Questions researchers ask.
Is there academic pricing?
Can I cite the sources in a paper?
I don’t write code. Can I still use it?
Enterprise on Valyu.
Private indexes on your firm's internal data, enterprise controls, and engineers who ship with you.
Talk to our team
SOC 2, Zero Data Retention, SSO with SAML or OIDC, scoped tenant isolation. US, EU, and UK deployment options, or dedicated in your VPC.
State-of-the-art retrieval on internal data: lab notebooks, internal reports, trial data, research archives. Agents run on your own research and every Valyu source, in one call.
When the API needs to go further, our engineers ship with you. Custom DeepResearch agents and vertical indexes tuned to your field.
One line to start
Paste this. Ask your first question.
Add Valyu to Claude Code and search the full-text literature from your terminal.
claude mcp add --transport http valyu https://mcp.valyu.ai/mcp