visual-collaboration

Streams agent-owned displays to a dashboard for live visual collaboration.

14|Updated Mar 2, 2026
One-click install
npx skills add https://github.com/intendant-dev/Intendant --skill visual-collaboration-intendant-dev
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: visual-collaboration
Source: https://github.com/intendant-dev/Intendant/tree/main/skills/visual-collaboration
Command: npx skills add https://github.com/intendant-dev/Intendant --skill visual-collaboration-intendant-dev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When an agent performs GUI work on its own display, the user has no visibility into what is happening. This Skill opens a shared view of the agent-owned display in the Intendant dashboard so the user can watch live work, review finished results, and take input authority when needed. ## Core Features & Use Cases - Shared Display Streaming: Create a virtual display (Xvfb on Linux) and stream it live to the dashboard with show_shared_view. - Focus Annotations: Highlight normalized screen regions with focus_shared_view and capture frames with capture_shared_view_frame. - Input Handoff: Ask the user to take input authority via request_shared_view_input for logins, passwords, or account pickers. - Use Case: After an agent finishes configuring an app's UI, it opens the shared view so the user can inspect the result, then highlights the relevant settings panel and hides the view when done. ## Quick Start Ask the agent to show you its display so you can watch the GUI work live in the Intendant dashboard.

Frequently Asked Questions about visual-collaboration

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I show an agent's display to the user in Intendant?▼

Call show_shared_view to open the shared display surface and mark the display as the shared view, which requests display-stream activation so the dashboard shows it live. If no agent-owned display exists, create one first with create_virtual_display.

How do I let the user type a password during agent GUI work?▼

Show the display and call request_shared_view_input, which asks the user to take input authority. Input authority is only granted when the user clicks the dashboard control; the tool never grants it directly.

Does create_virtual_display work on macOS or Windows?▼

No. create_virtual_display uses Xvfb and currently supports Linux hosts only; other platforms return a clear error. The shared view tools themselves work with any display the daemon can stream.

Can I share the user's own desktop screen with the shared view?▼

Only when the user has explicitly opted in by asking you to work on or look at their desktop, using the user_session display target. Never default to it or treat auto-detection landing there as consent.

Why does the visual-collaboration skill say the daemon appears down?▼

If the Intendant CLI resolves but the daemon does not answer, the daemon is not running and the skill stops with that message. If no CLI resolves at all, Intendant is not installed and the skill does not apply.