Mantis
v0.3.7

Extract pages and selected text as Markdown for agents.

Click Mantis in the browser toolbar. It reads the page already rendered in the active tab, converts it locally, shows a preview, and copies the Markdown to your clipboard.

Page or selection → Markdown → clipboard.

This browser tool is built using Mantis, a small extraction library. Read more →

Fig. 1 — click Mantis on a rendered page; the Markdown appears and is copied.

Install

Chrome Web Store — not published
Deferred

The Manifest V3 build passes local tests. A public install link will be added after store submission.

Before / after

Fig. 2 — page chrome is removed; useful structure and source details remain.

Capabilities

Selection first

A useful text selection takes priority over the full page.

Structure kept

Headings, links, lists, code, images, and tables stay in reading order.

Capture details

Source, mode, confidence, warnings, and token estimate remain visible.

Clipboard ready

Copies automatically, with a manual control when the browser blocks it.

Permissions

No persistent site access.

activeTab Read the tab you click Mantis on.
scripting Run the packaged extractor in that tab.
clipboardWrite Copy the Markdown result.

No external page fallback. Page URLs and content are not sent to a proxy service.

Limits

Known browser and extraction boundaries.

Restricted URLs Browser-internal pages, add-on stores, and other protected URLs cannot be captured.
Page state Hidden, unloaded, or inaccessible frame content is outside the rendered DOM snapshot.
Extraction Unusual layouts can produce partial or low-confidence output. Mantis is not a verbatim archive.
Images Image references and alt text are kept. Mantis does not run OCR or image analysis.