Pinned Tweet
Introducing agent-desktop
Computer use for your agents
▲ E2E test desktop apps with agents
▲ Reads apps via the accessibility tree
▲ Drills into part of a window to save tokens
▲ Runs many agents on one app at once
▲ Works on macOS, Windows & Linux
npm i -g agent-desktop
Computer use for your agents
▲ E2E test desktop apps with agents
▲ Reads apps via the accessibility tree
▲ Drills into part of a window to save tokens
▲ Runs many agents on one app at once
▲ Works on macOS, Windows & Linux
npm i -g agent-desktop
82
75
1,544
152,295
Lahfir ▲ retweeted
I benchmarked 5 computer-use tools
- agent-desktop
- Open Interpreter
- UI-TARS
- Cua
- OmniParser
The task: 40 real desktop tasks. Forms, icons, tiny buttons, a calculator and a text editor.
Same Ubuntu desktop, same tasks, checked against what actually changed on screen
- agent-desktop
- Open Interpreter
- UI-TARS
- Cua
- OmniParser
The task: 40 real desktop tasks. Forms, icons, tiny buttons, a calculator and a text editor.
Same Ubuntu desktop, same tasks, checked against what actually changed on screen
23
16
337
22,560
In the latest release of agent-desktop, your agents reliably operate macOS system surfaces: the Dock, the menu bar, and Control Center.
Also in 1.1.0:
• Refs stay on the exact element, even after the UI changes
• Visible typing caret and steady-paced typing
npm i -g agent-desktop
Also in 1.1.0:
• Refs stay on the exact element, even after the UI changes
• Visible typing caret and steady-paced typing
npm i -g agent-desktop
Introducing agent-desktop
Computer use for your agents
▲ E2E test desktop apps with agents
▲ Reads apps via the accessibility tree
▲ Drills into part of a window to save tokens
▲ Runs many agents on one app at once
▲ Works on macOS, Windows & Linux
npm i -g agent-desktop
19
11
331
16,674
WebMCP will solve a lot of technical problems in the web surface. canvases, iframes, custom renderers can expose tools for agents to interact instead of pixel matching.
wterm now supports WebMCP
Agents can use the terminals on your page
→ List, read, type, press keys
→ Opt-in per terminal, read-only mode
→ React, Vue, Svelte + vanilla
→ No DOM scraping or fake key events
Below: 5 fx agents launch each other across 4 browser terminals
3
0
29
2,030
Lahfir ▲ retweeted
What shipped at Vercel Labs in October
95 releases
730 issues closed
3,500+ PRs merged
New: agent-desktop
→ Computer use for any coding agent
→ macOS, Windows + Linux
New: tsc-ts
→ The TypeScript 7 compiler, written in TypeScript
→ Compiled to native with scriptc
agent-browser
→ Safer MCP
→ 42 browser tools for eve agents
skills
→ Sync skills across a monorepo
→ Sync reads optional + peer deps
→ Faster updates
json-render
→ Streaming fixes
→ Form + PDF fixes
→ json-render-shadcn skill
portless
→ No more sudo
→ portless mcp
deepsec
→ Kimi K3.1 Pro + DeepSeek V4.1 Flash benchmarks
native
→ native test for TS apps
→ Reusable markup components
→ Virtual lists
scriptc
→ Fully native, no JS engine
→ WebAssembly compiles to native
→ Profile-guided optimization
just-bash
→ mktemp + yes
→ Closer to real bash + jq
fx
→ Launches 23× faster
→ 41% fewer tokens
→ Agents that survive redeploys
wterm
→ Faster color rendering
→ Emoji + flags in one cell
→ Vim matches your page theme
vgpu
→ Build-time WGSL shaders
→ Turbopack support
→ Depth + MSAA on canvas
lat․md
→ lat ui to browse your knowledge graph
→ Graph view + static export
emulate
→ npx emulate mcp
→ AWS Cognito
→ Slack interactivity
ai-cli
→ ai mcp
→ ai decide can judge images
mcp-handler
→ MCP Events
run
→ CPU time limits for agent code
95 releases
730 issues closed
3,500+ PRs merged
New: agent-desktop
→ Computer use for any coding agent
→ macOS, Windows + Linux
New: tsc-ts
→ The TypeScript 7 compiler, written in TypeScript
→ Compiled to native with scriptc
agent-browser
→ Safer MCP
→ 42 browser tools for eve agents
skills
→ Sync skills across a monorepo
→ Sync reads optional + peer deps
→ Faster updates
json-render
→ Streaming fixes
→ Form + PDF fixes
→ json-render-shadcn skill
portless
→ No more sudo
→ portless mcp
deepsec
→ Kimi K3.1 Pro + DeepSeek V4.1 Flash benchmarks
native
→ native test for TS apps
→ Reusable markup components
→ Virtual lists
scriptc
→ Fully native, no JS engine
→ WebAssembly compiles to native
→ Profile-guided optimization
just-bash
→ mktemp + yes
→ Closer to real bash + jq
fx
→ Launches 23× faster
→ 41% fewer tokens
→ Agents that survive redeploys
wterm
→ Faster color rendering
→ Emoji + flags in one cell
→ Vim matches your page theme
vgpu
→ Build-time WGSL shaders
→ Turbopack support
→ Depth + MSAA on canvas
lat․md
→ lat ui to browse your knowledge graph
→ Graph view + static export
emulate
→ npx emulate mcp
→ AWS Cognito
→ Slack interactivity
ai-cli
→ ai mcp
→ ai decide can judge images
mcp-handler
→ MCP Events
run
→ CPU time limits for agent code
71
18
502
28,832
Lahfir ▲ retweeted
if you're on a mac and want computer use without the agent touching your personal stuff, here's how i do it
run agent-desktop inside a macOS VM. Apple's Virtualization framework boots a full guest in a window, and the agent only sees what's inside it.
you keep working on your s inside the VM
run agent-desktop inside a macOS VM. Apple's Virtualization framework boots a full guest in a window, and the agent only sees what's inside it.
you keep working on your s inside the VM
89
125
2,709
193,015
if you're on a mac and want computer use without the agent touching your personal stuff, here's how i do it
run agent-desktop inside a macOS VM. Apple's Virtualization framework boots a full guest in a window, and the agent only sees what's inside it.
you keep working on your s inside the VM
run agent-desktop inside a macOS VM. Apple's Virtualization framework boots a full guest in a window, and the agent only sees what's inside it.
you keep working on your s inside the VM
89
125
2,709
193,015
Lahfir ▲ retweeted
Get onboarded with agent-desktop 😮💨
Introducing agent-desktop
Computer use for your agents
▲ E2E test desktop apps with agents
▲ Reads apps via the accessibility tree
▲ Drills into part of a window to save tokens
▲ Runs many agents on one app at once
▲ Works on macOS, Windows & Linux
npm i -g agent-desktop
5
3
127
7,963
asked my fx agent to draw something for the halloween
this is the prompt i used:
"using agent-desktop cursor enabled multi agent, draw something for the halloween in the freeform app."
video not sped up
this is the prompt i used:
"using agent-desktop cursor enabled multi agent, draw something for the halloween in the freeform app."
video not sped up
Introducing agent-desktop
Computer use for your agents
▲ E2E test desktop apps with agents
▲ Reads apps via the accessibility tree
▲ Drills into part of a window to save tokens
▲ Runs many agents on one app at once
▲ Works on macOS, Windows & Linux
npm i -g agent-desktop
5
2
54
3,550
Lahfir ▲ retweeted
new computer use just dropped!
Agent Desktop has passed one of my most difficult computer use benchmarks.
plan + order my groceries.
Did this 100% in the background WITHOUT stealing focus from my cursor or my monitor.
Excellent work team.
Agent Desktop has passed one of my most difficult computer use benchmarks.
plan + order my groceries.
Did this 100% in the background WITHOUT stealing focus from my cursor or my monitor.
Excellent work team.
Introducing agent-desktop
agent-browser, for your whole computer
→ Computer Use for any coding agent
→ macOS, Windows and Linux
→ Runs in bg while you keep working
→ Agents use your computer in parallel, each with its own cursor
10
12
179
14,091