Does installing Surf start sending my browsing to a model?

No. Default Surf commands are local-only and never call the SDK. Semantic mode is opt-in per run with --allow-semantic, which enables bounded TypeSafe decisions for semantic.find and related commands and sends bounded page observations to TypeSafe. Its credentials live under separate XDG typesafe paths from the rest of Surf.

Is this another browser MCP server I have to configure?

Not quite. It is a plain CLI over a Unix socket: install the npm package, load the extension unpacked in Chrome, then run `surf install ` for the native host. No MCP server, relay process or subscription, and no provider lock-in, since anything that can run commands can drive it. Chrome, Chromium, Brave, Edge, Arc and Helium are supported, plus a WSL2 path to a Windows-side browser.

Can the browser run on one machine and the CLI on another?

Yes, over Tailscale, with per-client credentials. `surf remote authorize` writes a mode-0600 credential holding the client's Ed25519 private identity, which you move over a secure channel; give each client its own, revoke immediately if lost, and note that Tailnet reachability is not authorization. There is no insecure mode, downgrade, or plaintext retry, and remote listeners require a POSIX browser host.

Where exactly does Jev sit in the loop?

Per the documented sequence: Surf observes the page, Jev selects the next action, Surf validates and acts, then Jev verifies goal completion. Surf also resizes screenshots to 1200px automatically and captures one after each action to save tokens and round-trips, and it falls back gracefully when CDP fails. The README does not document which questions the semantic loop asks.

How do I install it and connect it to Chrome?

Run npm install -g surf-cli, load the unpacked extension from the path printed by surf extension-path, then run surf install and restart Chrome before testing with surf tab.list. Chrome is the default, and --browser selects brave, helium, edge, arc or chromium, or use --browser all.

Can it use my existing chatbot logins instead of API keys?

Yes. It can query ChatGPT, Gemini, Perplexity and Grok using your existing browser logins with no API keys needed. Smart defaults also resize screenshots to 1200px to save tokens, auto-capture a screenshot after each action to save round-trips, and warn rather than fail on restricted pages.

What does it capture on the network?

It automatically logs all network requests while active, so you can filter, search and replay API calls without manually setting up request interception.

CommunityJev Tools & CLIs

Surf

A CLI browser automation tool with optional Jev semantic mode: Surf observes, Jev selects the next action, Surf validates and acts, Jev checks the goal.

See highlights

Open github.com

Stars: 634
Language: TypeScript
License: MIT
Published by: Community
Author: nicobailon
Last verified: Oct 2026

nicobailon/surf-cli

User repository on GitHub

View on GitHub

The CLI for AI agents to control Chrome. Zero config, agent-agnostic, battle-tested.

Forks: 59
Last pushed: Updated Oct 2026

Repo stats from the GitHub API, cached Oct 2026.

About this project

The CLI for AI agents to control Chrome. Zero config, agent-agnostic, battle-tested. Browser automation for AI agents is harder than it looks. Most tools require complex setup, tie you to specific AI providers, or break on real-world pages. Agent-Agnostic - Pure CLI commands over Unix socket. Works with Claude Code, GPT, Gemini, Cursor, custom agents, shell scripts - anything that can run commands. Zero Config - Install the extension, run commands. No MCP servers to configure, no relay processes, no subscriptions.

Quoted from the project's README. Read the full README.

Tagscommunity browser cli computer-use

Works forJev for Realtime control Jev for Verification Jev for Routing

Used inJev for Software & DevTools

Languages: TypeScript
Platforms: Browser CLI
Capabilities: Open sourceRequires API key

Highlights

--allow-semantic opts into bounded TypeSafe decisions for semantic.find and related commands.
Documented loop: observe page, Jev selects, Surf validates plus acts, Jev verifies goal completion.
Credentials store under XDG typesafe paths separate from non-semantic Surf commands that never call the SDK.

Watch out

Semantic mode sends bounded page observations to TypeSafe; default Surf commands stay local-only.

Questions about this entry

Short answers drawn from the linked project or page.

Reactions & coverage

Posts, threads, and videos about this entry from around the web.

X: Jev driving real Chrome from Grok Bot

X: Headless Chromium agent

X: Computer use at 155× cheaper than Opus 5

Hacker News: Computer use on a Mac for 1/50th of a cent per step

rahimnathwani82 points · 60 commentsTypesafe-computer-use drives a Mac toward a goal for 1/50th of a cent per stepRead the thread on Hacker News

More Jev resources like this

jev-opus

Claude Code plugin and CLI that runs Claude Code on Opus 5.5 and has Jev re-decide the reasoning effort when a prompt arrives and after every tool batch, sending each change as a per-message effort statement inside one unbroken prompt cache.

Jev Tools & CLIs#community#claude-code#routing

Community

WXK-AI/jev-opusTypeScript

Sedum

Open-source, pre-alpha CLI that runs plain-English browser tests on Playwright, using Jev for bounded decisions: resolving each step, picking elements among offered candidates and judging verify claims with probabilities. Bring your own TypeSafe key; MIT licensed.

Jev Tools & CLIs#community#e2e-testing#playwright

Community

sedum-dev/sedumTypeScript

decision-router

A CLI, Claude Code plugin and Pi extension that pick a model per task from one Jev call: a Choice over candidates asked in both orders plus a complexity Score that sets a minimum tier, with quota-aware filtering and fallback.

Jev Tools & CLIs#community#routing#claude-code

Community

Ruivalim/decision-routerTypeScript

Cua

29.3k

The Cua computer-use platform documents a jev-use recipe where TypeSafe Jev picks bounded action IDs and Cua Driver executes clicks and fills in the browser.

Jev Demos & Experiments#community#browser#computer-use

Community

trycua/cuaRust

jcode

20.4k

Rust coding-agent harness (MIT, ~20k stars) whose crates/jcode-base/src/jev.rs is a typed TypeSafe Jev decision transport, used for memory pruning (memory_jev.rs), browser navigation (browser_jev.rs), and voice routing (voice_intent.rs) with separate provider selectors per purpose.

Jev Tools & CLIs#community#rust#coding-agent

Community

1jehuang/jcodeRust

jevgrep (code search)

2.6k

A CLI that asks Jev which files and declarations answer a repository question, then prints verbatim excerpts for a coding agent to use.

Jev Tools & CLIs#community#typescript#cli

Community

dzhng/jevgrepTypeScript

From the Jev guides

Original write-ups on what this entry is for.

Browser Agents and Computer Use: What to Copy from the First Breakout Category

Sep 22, 2026

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

Jev on the Gateways: Vercel, Workers AI, and OpenRouter

Oct 12, 2026

The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.

Jev Tools & CLIs#gateway#vercel#cloudflare

Read article

Exporting Decision Models: GGUF, ONNX, and Device Runtimes

Oct 10, 2026

Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.

Jev Tools & CLIs#open-models#local#onnx

Read article

Back to all resources

From the Jev community

Posts from builders shipping with Jev right now.

Follow @typesafeai