Changelog

Model picker moves to where you type

The model picker moved out of the top header and into the message bar — open the reasoning button to search and switch models without leaving the composer.

Model picker moved from the top header into the composer reasoning menu as a searchable submenu; header picker hidden on mobile, selection switches the chat model in place.

Draw on images, catch trending topics

Sketch directly on an attached image before sending it, and tap live trending topics under the composer to search what is happening now in your country.

Attached images open an editor with pen markup, undo and redo, with the edited image swapped into the attachment; trending topics appear per country and trigger a web search on tap.

A canvas for writing: essays, letters, emails

Ask VoidAI to write something you keep — an essay, story, letter, or email — and it lands in a dedicated canvas with a recipient line for emails and fill-in fields for names and details.

New writing canvas with a title, markdown body, optional recipient line, and bracketed placeholders that render as fillable fields.

VoidAI now sees your workspace images

Ask about an image saved in your workspace and VoidAI describes what is actually in it — diagrams, screenshots, photos. Text-only models say so instead of guessing.

Workspace image reads attach the picture for vision-capable models; text-only models receive an explicit note that the image was omitted.

CLI 0.1.1: safer logins for scripts and automation

Pipe your token with voidnet login --with-token so it never lands in shell history, unknown flags now fail with usage instead of being ignored, and logout keeps your local token when the server cannot be reached.

0.1.1 line: token intake over stdin for headless runs, strict flag parsing, and a logout that preserves local credentials on failed server revoke.

Compatibility promise: 30-day notice on breaking changes

Breaking changes to publisher-facing APIs are announced in this changelog at least 30 days in advance, with the current version kept routing until the window closes. Non-breaking fixes need no notice.

Standing policy per Publisher Agreement section 6. Full text lives there; this entry is the dated pointer publishers can reference.

Docs pages now serve plain markdown for agents

Every docs page has a matching .md URL, so AI assistants read clean markdown instead of scraping pages.

Each docs page serves a canonical markdown variant alongside HTML, plus llms.txt, robots, and sitemap entries.

Voidnet CLI 0.1.0: your terminal for Voidnet

Install with npm install -g openvoidnet. Script everything with JSON output, check service health, stay updated automatically — and sign in from your browser, with per-app machine tokens managed in Console.

First 0.1.0 line as a single binary: JSON output, health checks, self-updates; browser login with a consent screen; machine tokens bound to one app, managed from Console tokens; buyer and publisher guides in docs.

Wallet payments apply exactly once

Retrying a wallet top-up or grant can no longer charge or credit twice — a retry finishes the same payment instead of duplicating it.

Wallet grants complete as a single unit; retries finish the same grant instead of creating a second one.

Model picker tidied: DeepSeek Flash leads

Retired model names are gone and Groq chat models leave the picker — DeepSeek Flash is the clear default, and title generation keeps working as before.

Retired names collapsed to deepseek-flash; Groq chat models removed from the picker with title generation unaffected.

Publisher API v1 is live

Publisher API v1 is available today: manage API keys, read usage, and refresh tokens from code or the new Console tokens page. Domain verification now shows exactly what to prove, step by step.

Publisher v1 endpoints for keys, metering, and refresh; Console tokens page with per-app usage; domain verification walks through evidence in the UI.

Composer rebuilt: the clunk is gone

The composer used to lag and stutter — typing, drafts, attachments, and edits are instant now, tooltips behave, and tapping an attachment opens it in the workspace.

Composer and message bubbles rebuilt: instant typing and drafts, stable rows, tap-to-open attachments, focus-on-edit, and proper tooltips.

Tell us what's wrong, right where it happens

Every header now carries a feedback button — in chat, on the site, in Console, across docs, even on shared links. Tap it, describe the problem in your own words, send. No account needed, and the page you're on travels with the report so the team sees exactly what you saw.

The same dialog in all five apps; reports arrive with the page attached and work signed in or anonymously.

Real Word documents with charts, plus bigger reads

Ask for a document and get a real paged .docx with charts and a live preview — and VoidAI now reads files up to 15000 lines with longer answers.

Documents build as real .docx with charts and paged preview plus share previews; reads go to 15000 lines and 2MB with longer turns.

A calmer shell, sources you can trust, tidier images

The sidebar and drawers glide instead of jumping, the header gets out of your way while you scroll, search answers gather their citations into one Sources panel, and grouped images open in a cleaner viewer.

Sidebar and drawer motion rebuilt with scroll fades for header and dock; answers cite grounds in a Sources panel; grouped images open in an improved lightbox; signed-out users get a clear sign-in hint.

Newsroom shows edit dates, new VoidAI guides

Articles now show when they were last edited, alongside a rewritten VoidAI guide and a new workspaces explainer with screenshots.

List and article pages render last-edited dates; two guides added covering what VoidAI does and how workspaces organize projects, documents, notes, mindmaps, and files.

Shared mindmaps render cleaner with follow-up prompts

Opening a shared mindmap link shows a polished view with a pill to ask VoidAI a follow-up.

Shared mindmap links open a polished view; an ask pill sends a follow-up to VoidAI.

Seasonal artwork in the header

The header celebrates the season with light and dark artwork for Diwali, Halloween, Day of the Dead, Oktoberfest, and Mid-Autumn Festival.

The header shows seasonal artwork with light and dark variants.

Attach files to chat

Send files alongside your message, see them as chips, keep them in drafts, and let the assistant read their text — including PDFs.

Attach files in the composer with preview chips; drafts keep your attachments; the assistant can read attached files including PDFs, and uploads are kept safe.

Images group neatly, capable models see them

Multiple images in an answer arrange into clean grouped layouts with a better viewer, and models that support vision receive them while text-only models keep a safe default.

Images in answers arrange into grouped layouts with an improved viewer; models with vision support receive attached images, others continue with text.

Tap a file name to open it

When the assistant mentions a workspace file, it appears as a tinted button that opens the file in its workspace view.

Mentions of workspace files appear as buttons that open the file; folders open where supported; unknown paths show a clear not-found instead of guessing.

Slides with outline, stage, and present mode

Build slide decks as a single file with an outline filmstrip, a fixed stage, and a chromeless present mode — shared links present the same way.

Decks save with drafts and warn when a slide overflows; present mode hides all chrome; shared slide links present identically.

Same viewers everywhere, faster share links, self-healing sync

Notes, mindmaps, PDFs, and pages look the same in the workspace and in shared links, share links open faster with proper error pages, and file sync recovers on its own when something is out of date.

Workspace and shared links use the same viewers and share controls; share links open faster with proper error pages; file sync recovers on its own when out of date; text and mindmap selections can quote back into chat.

Codespaces: a dev space with shell and editor

Open a code space with a terminal shell, file-aware commands, project switching, and syntax-highlighted editing in familiar themes.

Codespaces combine an IDE panel, a shell with history and shortcuts, workspace commands, and project switching; the editor shows syntax colors; shell access stays fenced for safety.

Workspace files: code, PDF, quotes, safer sync

Code opens with highlighting, PDFs get a real viewer, selections can quote back into chat, and saving and syncing hold up under edits from multiple places.

Code opens with highlighting and PDFs get a real viewer with quote-back; saving and syncing hold up under edits from multiple places.

VoidAI feels instant and installs like an app

Replies drip in live, the app boots faster with friendly empty states, greetings feel personal, and you can install it with offline support and short chat links.

Replies appear word by word as they are generated; lightweight loading placeholders hold the layout; install prompts, update banner, and offline queue ship with short chat links and realtime panels.

API reference generated from the live spec

The API reference is now generated per product area with corrected fees, streaming, hosts, and model listing — grouped the way the docs navigate.

Reference pages are regenerated from the live gateway description with protocol-grouped navigation and redirects, so they stay accurate.

Knowledge cards, 13 languages, and a calmer VoidAI

Ask about a person, place, or thing and get a glanceable card with the key facts, all tappable for follow-ups. VoidAI now replies fully in 13 top world languages with regional dialects, renders wide tables that fit your screen, draws diagrams, and searches the web only when freshness actually matters.

Identity questions get fact cards with highlights, quotes, related topics, and links. Wide tables shrink to fit any screen. Diagrams draw automatically. The Language setting changes the reply language everywhere. File tools work with workspace files and views. Web search runs only for time-sensitive questions or on request.

Wallet, weekly payouts, and exact money everywhere

Top up a wallet and pay per use, get paid out every week as a publisher, and see exact amounts down to fractions of a cent across wallet, earnings, and analytics. New paid tiers are 80/20.

Safe retry top-ups; one payout per publisher each week with leftover fractions rolling into the next one; exact amounts everywhere; paid traffic stays fast under load; slow apps time out gracefully; clearer per-app error reporting.

Default platform fee 20%

New paid tiers default to publishers keeping 80%. Existing apps keep their locked rate; founding publishers keep 85/15.

New paid tiers take a 20% platform share. Past earnings untouched; founding rate unchanged.

AI models now publishable like MCP apps

AI models can now be published inside Voidnet just like MCP apps. Find it in openvoidnet.com/console.

Publish an AI model from the console with live verification before it goes live; browse models alongside tools in the marketplace.

Share any answer as an image

Turn any AI reply into a polished shareable image card — preview it, download it, copy it, or send it straight to your apps.

Share cards render the question and answer in light and dark themes with per-platform sharing where supported.

Streamlined publishing with a single working place

Publish in one clear flow — create a draft, verify your server, and set pricing in the same place. Drafts stay private until you publish.

Create on /publish-apps/mcp then verify and set tier on the app's overview. Verification shows your tools before you publish.

Subscribed Apps in Console

Find all your subscribed apps in one place and jump straight to usage.

New Subscribed Apps page lists your purchases with links to per-app usage.

Instant answers in VoidAI

Get beautiful inline cards for words, translations, definitions, grammar, and math — right where you ask.

Ask about a word, a phrase, or a calculation and VoidAI responds with a rich card before the answer.

Branch and revisit conversations

Edit any message, flip between versions, or branch a discussion into a fresh chat.

Your edits create new paths you can navigate and branch, keeping every version without losing context.

A cleaner Store and flexible model choice

Discover apps in a refined Store and use your own keys for top providers.

The Store and Installed Apps are now clearer and faster, with by-your-own-keys for leading providers and automatic model discovery.

Documents that feel like real pages

Create paginated documents with live updates that keep your place.

Documents now paginate with margins and headers, and edits appear inline without jumping your cursor.

Smoother, faster chat

Everyday chats feel quicker and handle busy moments more elegantly.

Under-the-hood improvements make common chats faster and limits easier to understand.

More reliable files and sharing

Your files and shared links are now more predictable and polished.

File actions are clearer and safer, shared links remain available while shared, and viewing code and images is smoother.

A workspace built for making

Five dedicated views for your projects, documents, notes, mindmaps, and files.

Each view is crafted for its content, with protected system folders and a unified setup.

Stateless MCP 2026-07-28 — sessions, initialize, and Mcp-Session-Id removed

Every MCP request is now independent. No setup handshake, no sticky server — horizontally scalable and faster.

BREAKING: only protocol 2026-07-28 is supported. Sessions, the initialize handshake, Mcp-Session-Id, and GET/DELETE session endpoints have been removed. All MCP communication is now single POST, stateless, with per-request SSE (close = cancel). Legacy 2025-11-25 clients receive 400 UnsupportedProtocolVersion -32022.

MCP-Protocol-Version + Mcp-Method/Name required, HeaderMismatch handling

Every POST now validates MCP-Protocol-Version header ↔ _meta.protocolVersion (HeaderMismatch -32020, HTTP 400) and unsupported version → UnsupportedProtocolVersion -32022 with {supported:["2026-07-28"],requested}. Adds Mcp-Method/Mcp-Name headers, ignores legacy Mcp-Session-Id/Last-Event-ID, forwards x-mcp-header. Transport keeps 16MB cap and per-request SSE extract.

server/discover + stateless list cache

Clients can discover what a Void App supports before calling tools — and list results are now cached correctly per buyer.

Gateway answers server/discover: supportedVersions is the gateway truth ["2026-07-28"], capabilities are the publisher's, serverInfo is the gateway in _meta, instructions from the publisher. List results support ttlMs/cacheScope; private results are isolated per buyer.

Per-app revenue split — default 80/20 (was 70/30)

Publishers keep 80% by default. Negotiated deals can keep more on a single app without touching code.

BREAKING: revenue split is now per-app, configurable via Console. Default 80% publisher / 20% platform, locked at purchase time. Publisher and platform shares always add up to the full price exactly.

Interactive OAuth — Authorization Code + PKCE + Refresh Tokens

Claude Desktop, Cursor, and VS Code can now log in via the gateway — open a browser, approve, and get a short-lived token without pasting keys.

New authorization_code flow: single-use codes, short-lived, requires PKCE S256 (43-char base64url challenge), bound to client_id and redirect_uri with exact-match validation. New refresh_token flow: rotates on use and rejects replay. Metadata now advertises authorization_endpoint and S256. Existing client_credentials unchanged.

VoidAI deep-link prompts (?q=) with auto-send

Open VoidAI with a prompt in the URL and it sends automatically.

VoidAI now supports ?q= deep links with auto-send, with web fetching and single citation enforcement.

VoidAI multi-provider model selector + search images

Pick Groq, DeepSeek, or OpenAI — and see images with every search.

Multi-provider model routing and search images with per-result cards, lightbox, and grounded highlights.

v1-beta gateway is live

Voidnet is now routing requests. Publish and consume MCP tools today.

The POST /v1-beta/{adapter}/{username}/{appname} endpoint is live. Stateless MCP 2026-07-28: per-request POST only, close=cancel, no GET/DELETE sessions. Updated 2026-09-02: the llm adapter is now available alongside mcp — see AI models entry; a2a is not yet supported.

OAuth 2.0 authorization server

Built-in OAuth 2.0 authorization server at POST /oauth/token using the client_credentials grant. Exchange an API key for a short-lived JWT (RS256). Well-known metadata at /.well-known/oauth-authorization-server (RFC 8414), /.well-known/oauth-protected-resource (RFC 9728), and /.well-known/jwks.json. Extended 2026-08-20 with authorization_code + PKCE (S256) and refresh_token (see interactive OAuth entry).

VoidAI keys supported

VoidAI now works seamlessly with Voidnet.

Voidnet now accepts vai-sk- API keys alongside vnb-sk- for integrated workflows.

Scope enforcement

Protocol adapters enforce scope prefix. A mcp: scope allows all MCP methods; an empty scope (legacy keys) remains unrestricted.

Correct MCP session forwarding (historical — removed 2026-08-21)

Historical (pre-2026-07-28): initialize forwarded to publisher and returned real capabilities, notifications/initialized forwarded, auto-init before method calls. Removed 2026-08-21 with stateless break — sessions no longer exist; see stateless entry.

Security hardening

Voidnet is now harder to abuse.

Request size limits on all requests. Sensitive information kept out of logs. Publisher API keys encrypted at rest. Suspended buyer accounts rejected across all sign-in flows.

Stripe billing

Publishers can now charge for their apps.

Stripe webhook integration for payment processing. Subscription and one-time payment flows. Tiered pricing (free and paid) with automatic meter enforcement. Updated: split is now per-app — see billing entries.