Native-quality rendering
Renders every page straight from the engine's own image buffer with no intermediate conversion that blurs detail, with zoom, rotation, thumbnails, and high-DPI export.
Agent Documents· Lector
Lector is a document engine we built for native-quality rendering, compiled to run inside any browser. It renders pages pixel-perfect with no quality-degrading image conversion, and goes far beyond viewing: it reads text, fills forms, permanently redacts content, and cryptographically signs and encrypts documents. The same engine ships as a drop-in viewer or as headless building blocks.
Why Lector
Most web PDF tools wrap aging, second-hand engines: they download the whole file before showing a page, lose quality by turning each page into a compressed image, and can only view. Anything serious, filling a form, redacting sensitive data, or legally signing, pushes you to desktop software or expensive closed SDKs.
Lector collapses all of that into one engine that runs in any browser and can be driven by software or a person, with nothing leaving the page.
Capabilities
Renders every page straight from the engine's own image buffer with no intermediate conversion that blurs detail, with zoom, rotation, thumbnails, and high-DPI export.
Extracts exact text with per-character positioning, click-and-drag selection and copy, and full-document search with case, whole-word, and regex options.
Sixteen drawing tools, from highlight and ink to shapes, arrows, free text, and sticky notes, with threaded comments, undo and redo, and standard import and export so markup is portable.
Fills every PDF field type, populates and extracts data in batch, and runs Adobe-style auto-calculations and formatting in a sandboxed engine.
A mark-then-apply workflow that does not just draw a black box: it deletes the underlying content and regenerates the page, so redacted text cannot be recovered.
Cryptographically signs documents with a certificate and verifies existing signatures for integrity, certificate, and expiry, with trusted timestamps for long-term validity.
Applies AES-256 encryption with user and owner passwords and granular permission flags for print, modify, copy, and annotate.
Visually diffs two documents down to the word, and inserts, deletes, rotates, reorders, merges, splits, and flattens pages.
Built different
Lector runs a current, hardware-accelerated rendering engine, instead of the years-old builds most web PDF libraries carry, with their known rendering bugs.
Competitors turn each page into a compressed picture, introducing artifacts and double memory. Lector transfers the raw rendered bitmap with zero copies, preserving exact fidelity.
Signing, verification, and encryption are implemented over a real crypto library compiled into the engine, producing standards-compliant signatures and AES-256 entirely in the browser, where most viewers either lack this or punt it to a server.
A framework-agnostic core with a thirty-plugin system, plus thin official bindings for React, Vue, Svelte, and Web Components, each shipping a full viewer and headless hooks.
All engine work runs off the main thread, and everything, including signing and encryption, happens client-side with nothing leaving the browser. Permissions can gate print, export, signing, and redaction per document.
By the numbers
Agent-first
A document engine turns an agent from something that can only discuss a contract into something that can act on it. With Lector in a sandbox, an agent can open a contract, read and search its exact text, fill in a form, black out personal data, then sign and return it, all programmatically, with nothing leaving the sandbox.
See HQ running in your own Slack or Teams, on the operating system we built for agents.