The Vibium Roadmap, Explained

Vibium's roadmap is demand-driven: the team ships what users ask for, and a lot has already landed. After v1 in December 2025, the February 2026 (26.2) release pulled several originally-deferred features forward — network interception, tracing, cookies and storage, clock control, and the Python client all shipped — reaching 168 commands across six targets at near-parity. What remains ahead, per the official roadmap, is broader browser support beyond Chrome, built-in video recording, AI-powered locators (page.do(), page.check()) for natural-language element finding, and official Java and .NET clients. The longer-term vision frames Vibium as a "Sense → Think → Act" loop: the shipped Clicker is the Act layer, with a memory/planning layer (Cortex) and a passive recorder (Retina) as exploratory future work. This article maps what's done versus what's coming, so you can plan adoption around reality rather than hype.

What has Vibium already shipped?

Most of the early "future" list is already in your hands. The v1 release delivered the core engine; the 26.2 release rounded it out far beyond the original plan.

If you assumed network tracing or the Python client were still "coming," they're not — they're current. See what is Vibium for the full picture of the shipped product.

What's next on the roadmap?

Four items lead the forward-looking roadmap, prioritized by user demand rather than fixed dates.

Broader browser support

Vibium automates Chrome only today, via a pinned Chrome for Testing download. Firefox and Edge support are roadmap items — Chrome covers the large majority of use cases, and BiDi implementations differ across browsers, so each is incremental work. If cross-browser coverage is essential now, pair Vibium with another tool until this lands.

Video recording

Built-in screen recording of browser sessions is planned, exposing start/stop controls and producing video artifacts for failure debugging, demos, and audit trails. Today, screenshots and the new tracing feature cover most debugging needs; video adds richer artifacts. For current capture options, see take a full-page screenshot.

AI-powered locators

This is the headline future feature: natural-language finding and assertions.

# Roadmap (not yet shipped) — illustrative API
# vibe.do("click the login button")
# vibe.check("verify the dashboard loaded")

It is also the hardest item — it raises open questions about which vision model to use, latency, cost, and how to resolve ambiguity. Until it ships, Vibium's AI-native story is its MCP server and semantic finds by role, text, and label, which already let agents target elements by intent.

Java and .NET clients

Official Java and .NET clients are planned, gated on enterprise demand. A community .NET implementation already exists, but is not officially supported. For now, official coverage is Python and JS/TS.

The longer-term vision: Sense, Think, Act

Vibium's architecture is framed as a robotics-style control loop, and only one layer is built so far.

Cortex (a datastore that builds an "app map" and plans multi-step navigation) and Retina (a Chrome extension that records sessions) are deliberately deferred — the team is unsure they add value over an agent's existing context, and will build them only if users report needing persistent memory or session replay. Treat these as direction, not commitments.

How should the roadmap affect my decisions?

Adopt Vibium for what's shipped, and watch the roadmap for what would unblock you. The practical reading:

• Chrome + Python/JS + AI agents — adopt now; the shipped feature set is complete for this.
• Cross-browser or Java/.NET needs — wait or pair with another tool; these are roadmap, not released.
• Want natural-language locators — use MCP and semantic finds today; full AI locators are still ahead.

The roadmap's guiding principle — "build what's requested, not what we assume is needed" — means the order can shift with community feedback via GitHub and Discord. For a maturity assessment alongside this roadmap, read is Vibium production-ready, and to start building, try your first Vibium script (Python).