Chrome extension · version 1.3.0
1.3.0: Capture HTML tables, extract links, and copy reviewed Markdown before adding a source. See the release notes.
Review a web page, selection or HTML table and copy its Markdown immediately. Or combine web captures and supported local files into one editable context-pack.md, with up to 10 sources. Choose the capture or import action.
Your content is processed locally. Copy or download the built Markdown before closing or reloading: the editable workspace is temporary.
Workflow
Add selectable-text PDF, DOCX, Markdown or text files. Use Capture page or Capture selection for browser content; capture opens or focuses the current workspace. See toolbar, right-click, keyboard and direct-PDF options. Wait until the current capture, review, or conversion is finished before adding another source.
Review captured pages and selections before adding them. Inspect Markdown, metadata and warnings, then build to see token estimates. In 1.2.0, switch between editable Markdown and read-only Preview. Reopen the first-run guide through Help.
Open current pack opens a new empty workspace or returns to your current one without capturing. Reorder, remove, include or exclude sources. New pack clears the session after confirmation when it contains work.
Edit the final Markdown, then copy it or download context-pack.md only when you choose. In 1.2.0, Preview does not change the exported Markdown, load images, or execute HTML.
Combine a research PDF, a DOCX brief and a web page into one context-pack.md, with a separate section for each source. See an example Markdown pack.
1.3.0 · Released
Capture a table, extract links, or copy a reviewed capture without adding a source.
Review a page or selection, edit its Markdown and inspect Preview, then copy immediately. Add to pack and Build are optional. Direct-PDF review is unchanged. See the flow.
Choose a visible HTML table in review, or highlight inside one and right-click Capture selected table to review that whole table. Inspect Markdown and warnings, then copy it or add one Table source. No PDF tables or virtualized grids. See a real table review.
Collect visible page or selected-region links. Apply keyword, origin and duplicate filters, then copy URLs, copy Markdown or add one Links source. No linked pages are opened or fetched. See the link filters.
Trust
Supported documents and ordinary captured text are processed inside the packaged extension. A direct HTTPS PDF is requested only after you explicitly capture that active URL, then parsed locally.
Chrome grants temporary activeTab access when you invoke popup, right-click, or keyboard capture. Context Pack reads page content only after that explicit action.
Working documents and captured content are not written to browser storage. Closing or reloading always discards the editable session. Only the non-content first-run-guide preference persists; copying or downloading keeps only the exported Markdown, not the source list or session.
PDF, DOCX, capture, Markdown, and token-estimation code ships with the extension.
Material limits
Maximum retained sources in one context pack.
Hard limit per PDF or DOCX; best under 10 MB.
Maximum per PDF, with 60 PDF pages per pack.
Version 1.2.0 retains supported visible action labels, selected non-editable choices, and disabled/pressed/expanded states. Complex tables become explicit row/cell text with a warning when a Markdown grid would be misleading. Input values and hidden choices stay excluded; this is not a full dashboard replica or automatic PII redaction.
Release notes
FAQ
No. Supported file conversion and token estimation run inside the extension with packaged code.
No. It is a local estimate based on OpenAI o200k_base for the generated Markdown and may differ for other models or chat applications.
No. Chrome grants temporary access only to the tab where you invoke the extension. Context Pack reads that tab only after you request page or selection capture.
