Jev Ultrafast: Inside the Speculative Browser Agent

Explain and reason about how Jev Ultrafast drives a browser with one TypeSafe request per decision cycle — from DOM snapshotting through speculative operation/target fan-out to guarded execution — well enough to trace a bug, extend the action space, or evaluate a similar agent design.

5 sections · 16 lessons

Course outline

The core idea: one request, two decisions

  1. Why browser agents are usually slow
  2. Speculative fan-out: asking for the answer before you need it
  3. The action space vocabulary

Seeing the page: snapshot and indexing

  1. Atomic DOM snapshot (snapshot.js)
  2. Indexed elements and the numbered element table
  3. Freshness guards and why retries are dangerous

Deciding: the model layer

  1. Dynamic operation and target heads (model.py)
  2. The small text-generation LLM for TYPE_TEXT
  3. Keeping the model out of the execution layer

Acting: execution and verification

  1. Executing a target safely (browser.py)
  2. Waiting without wasting time
  3. Background tabs, focus emulation, and visible text
  4. DONE is a claim, not proof

Putting it together and measuring it

  1. Walking the loop end to end
  2. What the benchmarks actually show
  3. Extending the action space without breaking the guarantees

Start learning with Wondering