Context Pack by TryAIBlock

Convert web pages, selections, PDF, DOCX, Markdown, and text into one clean Markdown (.md) file — locally.

Context Pack by TryAIBlock converts web pages, selected text, PDF, DOCX, Markdown, and plain-text files into one clean, editable Markdown (.md) file — locally in Chrome. Use the result as AI context, research notes, or any Markdown workflow. DATA HANDLED LOCALLY Context Pack handles only the files you choose and browser content you explicitly ask it to capture. For page capture, a local safety preflight traverses main-frame DOM text and attributes and measures the length of current input and textarea values before cloning, so oversized or complex pages can be rejected safely. Form values and filtered content never enter review, retained source state, or export. A visible switch, checkbox, or radio may contribute only its bounded accessible label, non-value state such as on/off, checked, mixed, or selected, and whether it is actually disabled, including when it appears inside an omitted form. The retained capture can include readable text, page title, safe links, and a sanitized URL from that current tab. This data is used solely to build your context pack inside the extension. It is not uploaded, sold, shared, used for advertising or analytics, or made available for human review. The working session stays in memory and is discarded when the workspace is closed or reloaded. Only a non-content preference recording that the first-run guide was closed is stored locally. WHAT YOU CAN DO • Add selectable-text PDF, DOCX, Markdown, and UTF-8 plain-text files. • Capture the readable current page or a current main-frame selection after an explicit action. • Review the captured title, sanitized source URL, warnings, and editable output before adding anything. • Invoke the extension from another source tab to add it to the same open pack. • Reorder, include, exclude, or remove sources. • Build and edit one combined document. • Copy the result or download context-pack.md. • View local token estimates for each source and the complete pack. LOCAL-FIRST PRIVACY • No account, backend, sync, analytics, advertising, or tracking. • Files and captured content are processed with code packaged inside the extension. • Page access is temporary and begins only when you invoke the extension and choose Capture page or Capture selection. • No persistent host access, browsing-history permission, background capture, or persistent content script. • No remote code or runtime code/dependency downloads. • Copying and downloading happen only after you choose them. • One local preference remembers that the first-run guide was closed; it contains no source or workspace data. CURRENT LIMITS • Up to 10 retained sources per pack. • Each .md, .markdown, or plain-text file must be valid UTF-8 and may be up to 1.5 MB. • Each PDF or DOCX works best under 10 MB and has a 20 MB hard limit. • Each PDF may contain up to 30 pages, with up to 60 PDF pages across one pack. • All retained sources in one pack may total up to 50 MB; this is a combined limit, not a per-file limit. • DOCX packages may expand to at most 80 MB during local processing. • The complete pack is limited to 500,000 UTF-16 code units. • Chrome 140 or newer is required. IMPORTANT LIMITATIONS • Scanned or image-only PDFs are not supported because OCR is not included. • Encrypted PDF and DOCX files are not supported. • Legacy .doc files are not supported. • Embedded DOCX images are omitted. Complex layouts and merged tables may lose fidelity. • Page capture uses the main frame only and removes forms, control values, interactive controls, embeds, media, images, navigation, advertising, and hidden content. A visible switch, checkbox, or radio may be represented only by a bounded plain-text label, non-value state, and actual disabled state. Action buttons, selected options, pressed/expanded state, and complex dashboard tables are not preserved reliably in version 1.0.0. • A local safety preflight transiently inspects main-frame DOM size and measures live form-value lengths; form values are omitted from review and export. • Browser-protected pages, iframe selections, text-field selections, and very large or complex captures may be rejected. • Token estimates use the packaged o200k_base tokenizer and may differ from other models or applications. • Context Pack does not import URLs, capture tabs automatically, or provide bulk capture. • Closing or reloading the workspace discards the editable session. Copy or download the result first if you need to keep it.