User repository on GitHub
Fast, cheap browser sub-tasks for Claude and other agents: TypeSafe Jev decisions on top of browser-harness
- Forks
- 1
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Integrations & Adapters
Jev Browse — danielnc
A Python toolkit that hands browser sub-tasks to Jev instead of the coding agent: one fast_run call lets Jev drive clicks and typing step by step, and the agent gets back a typed outcome or a typed reason to stop.
- Stars
- 9
- Language
- Python
- License
- MIT
- Published by
- Community
- Author
- danielnc
- Last verified
- Oct 2026
- Languages
- Python
- Capabilities
- Open sourceRequires API key
Highlights
- One decision equals one small Jev request of a few hundred milliseconds; no screenshots and no large model in the loop.
- A benchmark pair on Google Flights: 70.4 s and $1.41 driving browser-harness alone against 31.1 s and $0.45 through one fast_run call.
- Typed hand-backs cover the corners: confirm_required, in_frame, text_value_unavailable and more, alongside DONE and BLOCKED.
- Sensitive values are never read, and safety gates cover commit verbs, sensitive fields, frames, uploads and hosts.
- fast_run returns RunResult(status, reason, url, evidence, trace) and leaves a JEV_BROWSE_RESULT line to verify in the same script.
Watch out
MIT, a port of browser-use/jev-ultrafast. Needs Chrome, browser-harness connected to it, Python 3.11+ and a TypeSafe key; the 1.6-3.0x faster and 2.1-5.3x cheaper figures come from the author's own benchmark, and without the global pointer in your agent's instructions agents rarely reach for it.
Questions about this entry
Short answers drawn from the linked project or page.
Instead of the agent issuing click after click, each costing a model turn that re-reads its whole context, fast_run drives the whole sub-task in one call and returns only when it is done or must hand back.
The author benchmarked the same Claude Code Google Flights search on his machine, median of three: 70.4 seconds and $1.41 driving browser-harness itself against 31.1 seconds and $0.45 through fast_run, with benchmark medians of 72.8 s / $2.34 versus 24.7 s / $0.44.
From the values argument you pass, from the goal itself when Jev picks among candidates taken from it, or, only after a miss, from an optional text backend such as Claude, Codex, Ollama or any OpenAI-compatible endpoint, with a grounding gate before anything is typed.
On typed conditions: missing values, too many options, frames, shadow DOM, canvas, uploads, risky clicks or sensitive fields return reasons such as confirm_required, in_frame or text_value_unavailable instead of guessing.
Chrome, browser-harness connected to it, Python 3.11+, uv or pipx and a TypeSafe key: uv tool install jev-browse && jev-browse install, then store the key in the browser-harness agent-workspace .env and run jev-browse doctor until it passes.
Yes: /plugin marketplace add danielnc/jev-browse, /plugin install jev-browse@jev-browse, then /jev-browse:setup installs the package, wires browser-harness, and walks through the key, the text backend and doctor checks.
Yes, the README says agents rarely think to use it on their own: adding the global pointer to your agent instructions is part of the install, and doctor prints what is active afterwards.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Jev driving real Chrome from Grok Bot
X: jev-browser, a general browser skill
X: A second-hand shopping agent
X: A job crawler that reads careers pages with Jev
Other listings of this project
Forks and copies of this project in the directory. This entry is the canonical one.
4GitHub stars
Browser agent that takes a goal and start URL, reads each page as a numbered control table, and has Jev pick one operation and target per step until done or stuck.
Jev Demos & Experiments#community#awesomejev#catalog-import
Community
0x7067/jev-browseJavaScript
0x7067/jev-browseJavaScript3GitHub stars
Browser agent where Jev makes every navigation decision and agent-browser performs clicks and typing, stopping when the named goal page is reached or blocked.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
kyrylosyzonenko/jev-browseJavaScript
kyrylosyzonenko/jev-browseJavaScriptMore Jev resources like this
20.6kGitHub stars
Python agent framework whose TypeSafeModel class runs Jev as a decision model inside Pydantic AI agents: Pydantic output-type fields become typed Jev questions answered together in one request, each answer carrying its own confidence, and tools can be offered as Choice options.
Jev Integrations & Adapters#community#agents#framework
Community
pydantic/pydantic-aiPython
pydantic/pydantic-aiPython64kGitHub stars
Proof-of-concept multi-agent AI hedge fund that trades on paper: LLM investor agents analyse stocks, and a native JevLLM adapter lets an agent run on Jev's typed decisions through the same JSON contract the chat providers use. Educational only; no real trades.
Jev Integrations & Adapters#community#finance#trading
Community
virattt/ai-hedge-fundPython
virattt/ai-hedge-fundPython56GitHub stars
A UI testing framework where Jev scores candidate elements instead of generating next-step prose: natural-language YAML cases run through pytest and a browser agent, with Allure reports and a 458 ms median per decision.
Jev Tools & CLIs#community#python#testing
Community
buer2233/jev-ui-testPython
buer2233/jev-ui-testPython9GitHub stars
Epistemic runtime for autonomous coding agents that uses Jev as a sub-100ms System 1 gate to shadow-evaluate capsule facts, propose evictions before bloat and prefetch memory between turns. Its fixture run reports 31.3% fewer turns and 52.6% fewer blind discovery operations, with a fail-open circuit breaker.
Jev Integrations & Adapters#community#agents#context-engineering
Community
wojciechwiesner/jit-contextPython
wojciechwiesner/jit-contextPython7GitHub stars
Focus mode for Omarchy and Hyprland that judges each screen against the task you declared and escalates from a red reticle to a window haze and, as a last resort, closing the drifting browser tab. Each check is one Jev on-task classification, with screenshots OCR'd locally and never uploaded.
Jev Integrations & Adapters#community#focus#linux
Community
iYassr/omarchy-laserPython
iYassr/omarchy-laserPython187.6kGitHub stars
Autonomous agent platform with 187,504 stars for building and running agents from plain English or a visual builder. Its platform features first-class TypeSafe Jev decision blocks for typed routing, filtering, scoring and confidence-gated next-action dispatching.
Jev Integrations & Adapters#community#autonomous-agents#platform
Community
Significant-Gravitas/AutoGPTPython
Significant-Gravitas/AutoGPTPythonFrom the Jev guides
Original write-ups on what this entry is for.
Browser loops are long series of small decisions, and the expensive model does not need to make any of them. The patterns behind the fastest agents, from element tables to OCR plus accessibility trees.
Jev Tools & CLIs#browser#computer-use#agents
Read article
From the Jev community
Posts from builders shipping with Jev right now.