Pointframe — Claude Code MCP server: Windows screenshots and OCR
Gives agents access to Windows screen and window captures, text recognition, and video recording. Runs without installing a separate desktop app.
Open on GitHubWhat Pointframe MCP Server does
The server exposes eleven tools: monitor and window enumeration (list_displays, list_windows), screen and window capture (capture_monitor, capture_window), OCR (read_text_from_monitor, read_text_from_window), capture search (search_captures, get_capture), and recording (start_recording, stop_recording, get_recording_status). Each capture is saved as PNG with SHA-256, monitor geometry and DPI. Recording needs ffmpeg.exe in PATH.
Optional desktop_* tools automate UI tests: they launch an approved app, click controls, type text, and verify the interface, then write a signed report. Set them up with pointframe verify setup or a policy file for a specific app. The CLI and MCP server are standalone and need neither .NET nor the Pointframe desktop app.
What it helps with
- capture a monitor or window screenshot from the agent
- extract text from the screen with OCR
- record desktop video to MP4
- automate UI checks of a Windows app
- retrieve a saved capture by metadata
Example prompts for the agent
- Take a screenshot of the primary monitor and read the text from it
- Record the screen
\\.\DISPLAY1for ten seconds - Show the list of open windows with their titles
How to install Pointframe MCP Server
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Pointframe MCP Server
How to install Pointframe in Claude Code?
Clone the repository and run claude --plugin-dir ./plugin/pointframe from the root — this starts the MCP server and adds the verify-desktop-work skill.
What does Pointframe do?
It gives agents tools for Windows screen capture, text recognition, video recording, and automated UI verification of apps. The server runs without a separate desktop app.
Why does recording need ffmpeg?
The record command and start_recording tool use ffmpeg.exe to encode video to MP4. Without it in PATH, recording will not start.
