Movable Ideas / Nodes - Position to be remembered

Currently, the nodes are movable on the canvas, but its positions are not saved permanently. This is to be implemented ensuring the same position data will be rendered exactly as it should be on all the surfaces we support. Hence, consider the unit of measurement carefully for this implementation to remember the coordinates / position of nodes where they are. Also, a simple position movement would trigger Edit mode automatically ensuring 'Done' has to be pressed for the positions to be saved instead of constant saving.

Zero-Admin OS Launcher (Context Menu)

As a Visualli user, I want to right-click a local .pdf, .md or .txt file and choose Open with Visualli, so that my default browser opens https://alpha.visualli.ai and begins creating a Visualli (mindmap) from that file. This must work without installing a full desktop app or needing administrator rights. Scope note: This is a downloadable, cross-platform (Windows, macOS, Linux) shell-integration utility consisting of a user-space installer and a small native helper binary. It is not a browser extension. End-to-end flow 1. The user right-clicks one or more supported files and selects Open with Visualli. 2. The helper binary starts silently, validates the file type and size, and uploads the file(s) over HTTPS to a Visualli staging endpoint. The upload needs no login. The server returns a short-lived, single-use claim token (e.g. 15-minute TTL). 3. The helper opens the default browser to https://alpha.visualli.ai/import?token= and exits. 4. The web app checks authentication: - Signed in: it claims the staged files, attaches them as sources, and immediately begins generating the Visualli. - Not signed in: it redirects to sign-in/sign-up, preserving the claim token through the redirect. After authentication it returns to the import route, claims the files, and begins generation automatically with the originally selected sources. 5. If the token is expired or invalid, the web app shows a clear message and lets the user upload manually. Acceptance criteria 1. Zero-admin install: Installs entirely in user scope (Windows: HKCU registry and %LOCALAPPDATA%; macOS: ~/Library; Linux: ~/.local/share and ~/.local/bin). No UAC, sudo, or macOS Accessibility permissions. Binaries are code-signed (Windows Authenticode; macOS Developer ID signed and notarized; Linux checksums plus signed releases) to avoid SmartScreen/Gatekeeper warnings. 2. Native context menu: Entry labeled “Open with Visualli” appears only for .pdf, .md and .txt. - Windows: HKCU\Software\Classes\SystemFileAssociations\\shell\... (shown under “Show more options” on Windows 11 unless a packaged-app handler is added) - macOS: Finder Quick Action or Service in ~/Library/Services - Linux: per-file-manager integration (Nautilus scripts, Dolphin service menus, Thunar custom actions), installed in user directories 3. Execution and routing: The helper runs silently with no console window, uploads via HTTPS POST, then opens the default browser to the import URL. Multi-file selection is supported and results in one Visualli with multiple sources. It shows a native error notification if upload fails (offline, file too large, unsupported type). 4. Auth handoff: No credentials are stored or handled by the helper. Authentication happens only in the browser. Claim tokens are single-use, short-lived, and bound to the staged files. 5. Security and privacy: TLS only; the server enforces file-size limits, type validation (not just extension), rate limiting, and malware scanning; staged files auto-delete after TTL if unclaimed; the token travels in the URL but is useless after claim. 6. Clean uninstall: A zero-admin uninstaller (or --uninstall flag) removes all context menu entries, registry keys, services/workflows, and binaries, leaving no broken menu items. It is also exposed via Windows “Apps & features” and documented for macOS/Linux. 7. Distribution: Downloadable from the Visualli site (e.g. .exe/.msi per-user, .pkg or signed .dmg, and a Linux install script or AppImage), with versioned releases and an optional auto-update check.

Universal PDF & OCR Engine

As a Visualli user When I import a scanned document, image-based PDF, or non-native (non-text-encoded) PDF file, So that the backend automatically processes the visual pages using an OCR engine, allowing me to generate and augment visual maps from any PDF format just like Native PDFs (which it supports today!)

Map it! (Inside this map..)

Currently, 'Go Deeper' takes you to 'Map it!' which only creates a new visualli, however, the 'Inside this Map..' feature was not implemented yet. This would allow the map to be created from this point / node and the map grows further downward (in-place). so, the visualli is updated in-place to ensure it covers the entire map without overriding whatever is already in the children of that node.

Floating Layer Context Banner

Always know what you're looking at, no matter how deep you zoom. A subtle, floating context banner sits at the top of your canvas, giving a clear, high-level summary of the active layer (all the nodes) you are currently exploring. As you drill down or switch focus, the banner updates dynamically to explain the exact scope and theme of the visible nodes.

Edit Mode (Containers, Connections)

Currently, edit mode is only meant for ideas / nodes and its SemanticPeek text/summary. However, we want to use the design-system to implement the Edit mode for other available visual constructs as a part of the latest spec (0.2.6) which are Container Grouping, Container Title, Connections, Connection labels.

Visual Detail Lever

Control exactly how much information you see at a glance. A dynamic detail slider lets you tune your canvas density on demand - COMPACT | BALANCED | FULL - dial back to under 25% for a compact executive summary, set it to 50% for balanced everyday synthesis, or push above 80% for full, exhaustive coverage. Adjust your cognitive zoom level instantly based on whether you need a quick overview or deep technical depth.

Edit Visualli (Phase 1)

Move nodes position on the same layer, change text and labels, and make them persistent as you navigate deeper into layers. Ensure connecting lines remain properly anchored and adjusted when nodes are repositioned, maintaining visual integrity across the canvas.

Unified Canvas Color System

Maintain seamless visual harmony as you navigate. When you select a node, its color identity stays perfectly synced across the breadcrumb trail, and the background layer you are (about to) explore. By keeping these colors matched across every zoom level and detail view, you always have a clear, intuitive visual anchor for where you are in your research map.

Jump into Level-1 directly

Jump straight into your node structure with-out 'Home' or level-0 (single node name) filler. We are removing the blank default home container so your visual canvas opens directly on the first core layer of content—cutting out extra clicks and instantly surrounding you with meaningful information from the second you start.

Native Source Viewer

See your sources exactly as they were uploaded. Instead of converting original PDFs into raw text or Markdown - which can ruin visual formatting and cause confusion - Visualli opens the original document side-by-side with your map. View native PDF pages, layouts, and diagrams directly alongside your visual nodes so you can effortlessly trace insights back to their original context. RELATED FEATURE: Multi-span Citations

Browser Extensions (Chromium-based)

Transform how you read the web in real time. Install our browser extension to map out any article, thread, or documentation page on the fly across Chrome, Edge, Brave, or Arc. Visualli sits alongside your web browsing, rendering dense online text into spatial visual structures so you can absorb complex information naturally without breaking your flow.

Search across all Visuallis

Find any idea across your entire workspace in seconds. Enter a term or concept into the search bar to scan all your saved maps simultaneously. Results pinpoint the exact node and canvas where the phrase appears, allowing you to jump straight to that specific branch with a single click for instant recall and easy cross-referencing.

Explore — Interact with what matters

While navigating a Visualli, the information displayed hovering a node might not be enough. You might need more details or go deeper about it.Dig deeper into any node through conversation, exploration, or similar interactions. Follow the topics that matter to you, grow your Visualli, and shape your own field of knowledge.

Breadcrumb Sibling Reveal

Explore alternative node possibilities without losing your place. When you hover over any dot or topic in your top-left breadcrumb trail, a preview drawer that displays all neighboring nodes from that same layer as a horizontal drawer right next to that breadcrumb bubble, giving you a clear view of the surrounding node structure and availability and letting you jump directly across parallel nodes from different level in a single click.

Save Points

Never lose your train of thought while diving deep. As you explore down endless levels, layers and nodes, pin any specific view that you're seeing (e.g. specific level+layer+node) as a named save point at the top of your screen. When you need to anchor back to your starting idea or jump between key milestones, a single click restores that exact view, zoom level, and focus area—no matter how deep into the rabbit hole your research takes you.

Share visuallis (maps) with Friends

Share your exact visualli (map) view with friends using a simple auto-generated link. Friends can instantly explore your visual layout without logging in. Once they want to edit, customize, or use advanced features like Dig Deeper, Bring Anything they can save a copy straight to their own account with one-click to make it entirely their own.

Refresh a visualli

Notice when the content of a URL has changed, offer to update the visualli based on new content (and highlight changes?)

Align the colors of the index with the active sections

The background of the canvas changes with the selected item/blob. It would make sense to use that same color for the hierarchical index.

dark mode improvements