We will be undergoing planned maintenance on January 16th, 2026 at 1:00pm UTC. Please make sure to save your work.

Inspiration

Basically everything is running with AI models so would don't do it if we have the data, I got the inspiration from kaggle community on a project that i run in to.

What it does

  • Chat-driven assistant in VS Code that:
    • Answers “can I use this?” with Baseline Newly/Widely signals and dates.
    • Suggests and applies safe modernizations with inline rationale.
    • Discovers relevant features for your codebase and target browsers.
  • Uses a hybrid data model: local web-features for reliability + Web Status API for freshness.

How we built it

  • VS Code extension with commands (discoverFeatures, refactorModern, checkFeature, openChat) wired through a chat panel UI.
  • Data layer integrates the web-features package and real-time fetches (with caching/fallbacks).
  • Prompting and providers for Claude/Gemini/Ollama, tuned for “Baseline-first” reasoning.
  • Typed TS models for feature status, dates, and compatibility mapping.

Challenges we ran into

  • Normalizing different data shapes (package vs. API) and keeping queries fast.
  • Avoiding network brittleness while still surfacing the newest features.
  • Making NL→code edits trustworthy, explainable, and reversible in-editor.

Accomplishments that we're proud of

  • Seamless NL→code refactors backed by Baseline dates and status.
  • A reliable-offline, fresh-when-online pipeline that just works.
  • Clear, skimmable explanations that remove “can I ship this?” uncertainty.

What we learned

  • Developers want decisions, not data: Baseline signals must be actionable at the point of edit.
  • Caching + local-first drastically improves reliability without sacrificing recency.
  • Good prompts + feature context produce safer, smaller, more adoptable edits.

What's next for Baseline.dev

  • Deeper queries (groups/snapshots/date ranges) exposed as UI filters.
    • Based on baseline_status, group, baseline_date, snapshot from the Dashboard API docs: guide
  • Project-wide audits with autofix plans and PR-ready edits.
  • CI checks to block regressions against your target browsers.
  • Recipes for common modernizations (View Transitions, Popover, Container Queries), kept fresh via the Web Status API.

Built With

Share this project:

Updates