Urgent.News

What's breaking now, across thousands of outlets.

AI

Ultra Bridge: Translating JAWS Keyboard Commands to NVDA with Sanity Context

This is a submission for the Sanity Challenge, Path One: Ship an Agent That Queries Real Content What I Built Ultra Bridge is an agent for screen reader users. You ask it things like "What is the NVDA equivalent of JAWS Insert+F7?" or "Which keys do different jobs in JAWS and NVDA?", and it answers from a Sanity dataset through Sanity Context MCP. Every answer says how well the underlying command…

Ultra Bridge is an agent designed for screen reader users, allowing them to ask questions such as "What is the NVDA equivalent of JAWS Insert+F7?" or "Which keys do different jobs in JAWS and NVDA?" The agent retrieves answers from a Sanity dataset via Sanity Context MCP. It verifies the accuracy of each answer, enabling users to determine whether to trust the information or verify it themselves.

The author created Ultra Bridge due to the difficulties they faced when transitioning between JAWS and NVDA, as they found it challenging to adapt to new key commands. This challenge exemplifies the kind of issue structured content can efficiently address. The project is part of the Ultra suite of accessible software the author is developing.

Structured content is crucial for Ultra Bridge because while a keyword search might find a page mentioning both screen readers, it cannot provide specific information about key collisions, clean translations, or missing commands on one side but not the other. In Ultra Bridge, actions (like showing a list of links) are separate from commands (the specific key combinations a screen reader uses for those actions).

By using shared actions, relationship questions can be answered with a single query instead of guessing. The agent lists all keys used by both screen readers but assigned to different jobs. For example, the letter 'R' represents 'region' in JAWS and 'radio button' in NVDA. The agent returns six such keys in its initial query, highlighting the potential time savings and increased confidence for users when switching screen readers.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Your AI Vendor Just Became a Supply-Chain Risk

Your AI Vendor Just Became a Supply-Chain Risk A U.S. appeals court this week upheld a designation of a major AI company as a supply-chain risk . Read that phrase slowly.

  • AI vendor classified as supply-chain risk by U.S. appeals court
  • Companies face concentration, substitutability, and continuity risks
  • Mitigation strategies include abstracting layers, owning assets, and diversifying channels

Your LLM provider is probably serving you 32K context no matter what the model card says

I run a hosted chat and coding agent on open-weight models. This is the single finding that cost me the most time in the last few months, and almost nobody talks about it. The number on the model card is not the number you get The context window advertised on a model card is a property of the weights.

  • Hosted endpoints often serve 32K token context, not model card claims
  • Providers determine actual context window, not model weights
  • Context loss can be catastrophic for agentic tasks, invalidates RAG tuning

More from Saturday 26 September →