Jev.aitools.fyi

Command Palette

Search for a command to run...

Product profile
Swift663 starsUpdated 2026-09-24

TipTour

macOS menu-bar companion: default JEV mode picks local controls for clicks; Gemini mode for voice and writing. Keys in Keychain.

Overview

TipTour (github.com/milind-soni/tiptour-macos, MIT) is a macOS 14.2+ menu-bar app with two modes: Gemini realtime voice or screen writing, and JEV text mode (default) where you type a click task, Jev chooses among locally detected labels and coordinates, and TipTour executes single, double, or right clicks with validation. Ctrl+K starts Jev mode; Escape or Stop cancels. README documents a 12-action cap, no confidence cutoff on the top target, and keys stored per mode in Keychain without hosted proxies. Jev cannot see images or generate replacement text; Gemini handles keyboard and writing tasks. scripts/test-jev.sh runs isolated decision tests without launching the full app.

Problem: macOS computer-use demos that send screenshots to giant multimodal models are slow, costly, and hard to stop mid-loop.

Built for: Mac users who want a menu-bar agent that picks among locally detected controls with Jev, while keeping API keys in Keychain and screenshots local for the Jev path.

First indexed on Jev Directory: 2026-09-24

Creator and team

Name
Milind Soni

How Jev is used

Role in the product flow
Per-step Choice over locally detected control candidates for click automation
Primitives
Choice
State in
Typed user task, locally detected screen labels and locations, and recent action history per README privacy section (screenshots stay local for Jev).
Decision out
Ranked click target and action type; loop ends on completion signal, failure, user stop, or twelve actions.
  1. Grant Accessibility and Screen Recording permissions for detection
  2. User enters task in JEV text panel (Ctrl+K)
  3. Jev selects among detected controls; TipTour executes click
  4. Validate outcome and repeat until done, stop, or action limit

TipTour treats Jev as a fast click router over finite local candidates, not a vision model. awlevin-typesafe-computer-use pairs OCR plus accessibility with TypeSafe on macOS CLI loops; sac-y-jev-cu does text-only AX lists inside Codex. TipTour is a polished menu-bar product with branded shortcuts and a separate Gemini path for voice. Featured on community lists such as jevlist.ai TipTour coverage. Auto-click is required for Jev; point-only mode is Gemini-only per README.

Sourced performance claims

  • README lists JEV as default mode with 12-action limit and no confidence cutoff on the top-ranked target.Source: github.com/milind-soni/tiptour-macos README
  • Public GitHub repo milind-soni/tiptour-macos had 663 stars and 103 forks when this listing was drafted.Source: GitHub API September 2026
  • Jev path receives typed task and local labels only; screenshots stay local per privacy section.Source: github.com/milind-soni/tiptour-macos README

Features and stack

Features

  • Default JEV text mode with Ctrl+K shortcut
  • Single, double, and right click execution with stop controls
  • Per-mode API keys in Keychain; no shared hosted key proxy
  • Separate Gemini realtime mode for voice and writing
  • test-jev.sh harness for decision-only debugging

Stack

  • Swift
  • macOS 14.2+
  • Xcode
  • Jev API
  • Gemini realtime optional

Pricing: Open source app; you bring Jev and optional Gemini API keys.

FAQ

Does Jev mode send screenshots to the model?
README states Jev receives typed tasks and locally detected labels and locations while screenshots stay local. Gemini mode may send screenshots when enabled.
Can TipTour type text for me in Jev mode?
No. Jev mode supports click actions only. Use Gemini mode for keyboard or writing help.
How is this different from typesafe-computer-use?
Aaron Levin's CLI targets terminal-driven loops with dry-run defaults. TipTour is a menu-bar consumer app with shortcuts, dual models, and Keychain key storage.

Related learn guides

Original Jev guidance that pairs with this product pattern.

  • Jev use cases

    The patterns builders actually search for: moderation, routing, triage, RAG verify, and agent gates.

Related products

Hand-picked neighbors with rich profiles or overlapping tags.

Browser and computer usePython828
Demo
Computer use for about $0.

by @awlevin

browser-computer-usebrowser-usemacos
Browser and computer useJavaScript587
Codex Computer Use skill: Jev picks element, action, completion, and risk from AX text only; dry-run and confirm on sensitive ops.
browser-computer-usecodexcomputer-use
Browser and computer usePython4,799
DemoFeatured
Jev picks the browser op and DOM target in one shot; a tiny LLM only shows up when something needs typing.

by @gregpr07

browser computer usebrowser-useagents
By Rishit