Skip to main content
Glama

English · Русский

Step-by-step tutorials drawn right on top of any Windows app — led by an AI agent.

The lesson above was played through by the project's test robot in FL Studio.

Tutorial

To get started, install TutorialOverlay and ask your AI agent how to do something in an app:

TIP

“Walk me through turning on word wrap in Notepad, step by step”

The agent reads the app's interface over MCP and writes the lesson — steps, targets and a why for each one:

{
  "title": "Word wrap in Notepad",
  "app": "Notepad",
  "steps": [
    {
      "title": "Open the View menu",
      "why": "Notepad keeps everything about how text looks in the View menu.",
      "targets": [{ "name": "View", "anchor": { "element": { "name": "View", "role": "MenuItem" } } }]
    },
    {
      "title": "Close the menu with Esc",
      "until": [{ "type": "key", "keys": "Esc" }]
    }
  ]
}

The overlay rings each target on top of the app, the mascot points at it, and a step card leads the way. Steps complete by themselves — a click, a pixel, a window, a shortcut, a drag, a scroll. Stuck? Press Help and the agent answers right in the card. At the end the lesson can be saved and replayed from the tray, no agent needed.

For the full spec format see docs/tutorials.md.

Related MCP server: screencast

Installation

NOTE

Windows 10 or 11, x64. Nothing else: .NET is inside the exe.

TutorialOverlay is a local MCP server, so any agent that speaks MCP can drive it. Download TutorialOverlay-<version>-win-x64.zip from Releases, unzip it (say to C:\Tools\TutorialOverlay) and put it on PATH as tutorial-overlay:

Copy-Item C:\Tools\TutorialOverlay\TutorialOverlay.exe C:\Tools\TutorialOverlay\tutorial-overlay.exe
$p = [Environment]::GetEnvironmentVariable('Path', 'User')
[Environment]::SetEnvironmentVariable('Path', "$p;C:\Tools\TutorialOverlay", 'User')

Then, in Claude Code, install the plugin — the server plus the skill that teaches the agent to write good lessons:

/plugin marketplace add kizrum/tutorial_overlay
/plugin install tutorial-overlay@tutorial-overlay

That's all: no key, no port. The agent starts the overlay when it needs it, and it stays in the tray until you quit it from the tray menu.

TIP

Claude Code with the plugin works best. Other agents get the same tools and the server's short instructions, but not the skill.

Add to ~/.codex/config.toml (or .codex/config.toml in a project):

[mcp_servers.tutorial-overlay]
command = "tutorial-overlay"
args = ["--stdio"]

Add a local (stdio) MCP server: the command tutorial-overlay (or the full path to TutorialOverlay.exe) with the argument --stdio. In Claude Code:

claude mcp add --scope user tutorial-overlay -- tutorial-overlay --stdio

Do not add it this way next to the plugin: the agent would see every tool twice. Clients that speak Streamable HTTP can also connect to the running overlay directly — see docs/server.md.

Quit TutorialOverlay from the tray menu, unzip the new version over the old folder and copy TutorialOverlay.exe to tutorial-overlay.exe again. Settings, the key and saved tutorials live in %LOCALAPPDATA%\TutorialOverlay and stay as they are; the agent's settings stay right too.

Building from source

Needs the .NET 8 SDK. From the repository root:

# build: publish\TutorialOverlay.exe
dotnet publish -c Release -o publish

# run the tests
dotnet run --project tests

# use your build in Claude Code
claude mcp add --scope user tutorial-overlay -- "$PWD\publish\TutorialOverlay.exe" --stdio

.\release.ps1 builds the release zip. How the program is built inside is in docs/development.md.

Reference

  • Server — command-line options, the tray icon, settings, config.json, security, the log

  • MCP tools and events — what the agent can call and what it gets back

  • Tutorials — the spec format: shapes, anchors, steps and conditions, the mascot, help

  • Development — the inside of the program, the tests, the test robot

License

MIT


Shazhok says hi. Like everything else here, it is drawn in code — no images, no fonts, no packages.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables an AI agent to operate Windows applications through vision-driven UI Automation and record polished demo videos with pre-click camera zoom, narration, and cinematic effects.
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Enables users to drive a selected Windows application toward a natural-language goal by reading the UI with local OCR/UIA and deciding next actions.
      6
      5
      MIT
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI agents to read what is on a Windows desktop as text — open windows, their contents, and what changed — using Win32 and UI Automation with optional NPU vision. Secrets, password fields, and private windows are withheld and personal data scrubbed on the user's PC before anything reaches the agent.
      AGPL 3.0