Install from Chrome Web Store
Install from Edge Add-onsAgent
Connect Mdtero to your agent
OpenClaw, Claude Code, Codex, and Gemini CLI all start from the Account message with your setup already prepared.
Guide
Start in Account. Only open Chrome or Edge when the paper page itself needs local capture.
Most people only need one path: open Account for keyword discovery and API-key management, create a key, and send Mdtero to the agent they already use. Chrome or Edge is the secondary path when a live paper page has to be captured on your own machine.
Account is the normal setup path. The local helper handles acquisition by default, and Chrome or Edge is only for pages that must be captured in a browser.
Quick Start
Most people only need these two steps
Keep the default path simple: set up your agent in Account first, then install Chrome or Edge only if a live paper page must be captured locally.
Browser stores
Use Chrome or Edge only when you need local capture
- Open the Chrome Web Store or Edge Add-ons listing.
- Add Mdtero to the browser you already use and pin it if you capture often.
- Skip this entirely when Account plus agent handoff is enough.
Supported Sources
What works well right now
Most papers follow the same flow: start in Account or the local helper, and only bring in Chrome or Edge when a live publisher page must be captured locally.
Elsevier
Best for structured publisher XML, remote figure links, source downloads, and full parsing with your own developer key on a machine that already has access.
arXiv
Best for HTML-first preprints, faster starts, and carrying figures plus Markdown into your reading workflow.
Springer / Nature / RSC
Browser capture remains the fallback when a live publisher page needs to stay on your own machine first.
Downloads
What you get back
- Markdown
- humbert2023topology.md
- Embedded figures in humbert2023topology.md
- Translated Markdown
PDF is optional. Mdtero is designed to start from the structured Markdown file first, then fall back to source-side PDF or a ZIP only when a workflow truly needs it.
Good to know
A few cases worth knowing before you start
Most people can stop at Account and keep moving. These notes only matter when your reading workflow is a little more specific.
Is your OpenClaw running on a server?
Keep local or campus-bound acquisition on your own machine first, then pass the Markdown output back to OpenClaw for the rest of the workflow.
Do you need scripts or product integration?
Authentication, endpoints, and automation examples live on the API page so this guide can stay short.