Changelog
Shipping, in the open
Every release that makes feedback sharper and your agent faster. Newest first.
- v2.0NewJune 2026
The AI Context Engine
- Run an AI diagnosis on any annotation: ranked likely root causes, a confidence score, a suggested fix, and the source files it most likely lives in.
- Bring your own key for OpenAI, Anthropic, Gemini, OpenRouter, Mistral, or Groq. Keys are encrypted at rest and used only for your own requests.
- Every AI run is metered, so you always see how many tokens it used and what it cost.
- Prompt targets now include Gemini and Codex alongside Claude, Copilot, and Cursor.
- v1.9NewJune 2026
Share links & team notifications
- Send a read-only review link to anyone: stakeholders see the annotations and full context with no account, and reporter emails stay private.
- Revoke a link any time to instantly close access, or set it to expire on its own.
- A new in-app notification bell keeps your team in the loop the moment feedback arrives, a comment is posted, or a status changes.
- v1.8NewJune 2026
Subscriptions & autopay
- Plans can now renew automatically: UPI AutoPay, cards, and eMandate via Razorpay, plus PayPal subscriptions.
- A clear payment picker shows your autopay method, the next charge date, and any past-due retry.
- Cancel any time from billing; paid access runs to the end of your period, then drops to Free.
- v1.7ImprovedJune 2026
Account security
- Two-factor authentication (TOTP) with single-use backup codes, for any authenticator app.
- Email verification on sign-up and a self-service password reset over a secure, single-use link.
- Sensitive secrets are hashed or encrypted at rest, never stored in the clear.
- v1.6ImprovedJune 2026
A more capable MCP server
- Agents can now run an AI diagnosis, read attached screenshots as images, and browse comments and linked GitHub issues, all over MCP.
- A guided fix workflow walks an agent from an annotation through context, diagnosis, fix, and status update.
- Every tool is tagged read-only or write, so your client can auto-approve safe reads and confirm the rest.
- v1.5NewMay 2026
One-click GitHub issues & the Copilot coding agent
- Turn any annotation into a GitHub issue: the full captured context and an AI-ready prompt go straight into the body.
- Hand the issue to the GitHub Copilot coding agent in one click; it starts work and opens a pull request for you.
- Live issue history on every annotation: follow Open → Copilot working → PR in review → merged without leaving Vynix.
- Explore it all on the public, read-only live demo. No signup required.
- v1.4NewMay 2026
Runtime diagnostics capture
- Each note now carries recent console errors and failed network requests.
- Privacy-safe by design: metadata only, query strings stripped, opt-out via data-capture-diagnostics="false".
- Diagnostics surface in the Markdown report, the dashboard, and the generated AI prompt.
- v1.3ImprovedMay 2026
Batched feedback & “Copy all for AI”
- A round of feedback is grouped into a single batch across pages.
- One click copies a clean, page-grouped Markdown report for your agent.
- v1.2NewApril 2026
Region & text-selection capture
- Drag to capture a region, or select exact text for typos and copy edits.
- Freeze animations to annotate a single frame.
- v1.1NewMarch 2026
Model Context Protocol server
- Agents read annotations live and can resolve them in place.
- Works with Claude Code, Cursor, Copilot, Windsurf, Gemini and any MCP client.
- v1.0NewFebruary 2026
Vynix is live
- Click any element to leave structured feedback on a live site.
- Triage in the dashboard and generate an AI-ready prompt per annotation.
Stop describing bugs. Point at them.
Add one script tag and give your AI agent the context it has been missing. Free while we grow.