Skip to content

Jev directory

Argus

A source-available visual UI testing agent in Go: point it at a URL, describe the test, and a six-stage pipeline drives a real Playwright browser while Jev optionally validates requests, picks DOM targets and checks outcomes, falling back to an LLM below 0.9 confidence.
Stars
45
Language
Go
Reading time
3 min
Published by
Community
Author
ownweights
Last verified
Oct 2026

Highlights

  • With TYPESAFE_API_KEY set, Jev selects fresh element references and checks DOM/text outcomes; decisions under 0.9 confidence fall back to the selected LLM.
  • Jev calls have a 3-second timeout and a 64-call run budget; an API error disables Jev for the rest of that run.
  • Six-stage pipeline: validator, comprehender, explorer, strategist, executor, critic, with validated JSON handoffs and one repair attempt.
  • Source-available under ASAL-1.0: free under 100 members, commercial licence above; releases convert to MIT after 3 years.
  • Local-first Go server with SQLite storage and a stdio MCP adapter; 45 stars, created 2026-08-17, pushed 2026-10-03.

Quickstart

bash
cd frontend && npm ci && npm run build && cd ..
go -C argus run ./cmd/argus install-browser
GEMINI_API_KEY=your-key GEMINI_MODEL=gemini-2.5-flash ARGUS_RUN_TIMEOUT=300 ARGUS_DB_PATH=data/argus.db PORT=8000 go -C argus run ./cmd/argus

Watch out

README reviewed 2026-10-17; the digest excerpt truncates mid-sentence in the browser-capabilities list. Jev runs only when you configure a TypeSafe key, the 0.9 confidence gate is described as an initial gate rather than a measured accuracy guarantee, and the release has no server authentication or multi-user isolation, so the README says to bind it to a trusted interface.

Questions about this entry

Short answers drawn from the linked project or page.

From the Jev guides

Original write-ups on what this entry is for.

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai