gitwtfhub

wtf is anscribe?

msmps/anscribe — explained in plain English

Analysis updated 2026-05-18

14TypeScriptAudience · developerComplexity · 2/5LicenseSetup · easy

TL;DR

A tool that captures live terminal UI elements and sends them, with your instructions, to an AI coding assistant.

Mindmap

mindmap
  root((anscribe))
    What it does
      Captures terminal UI
      Sends to AI agent
    Modes
      Clipboard default
      MCP opt in
    Tech stack
      TypeScript
      Bun
      OpenTUI
      SQLite
    Packages
      opentui adapter
      mcp server
      react enrichment
      core

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Why would anyone build with this?

REASON 1

Send a screenshot-like capture of a terminal UI element straight into Claude Code or Cursor.

REASON 2

Give an AI agent automatic access to terminal UI state through a local MCP server.

REASON 3

Debug or improve an OpenTUI-based terminal app by capturing what it renders.

REASON 4

Share terminal UI context with an AI assistant over SSH without extra tooling.

What's in the stack?

TypeScriptBunOpenTUISQLiteMCP

How it stacks up

msmps/anscribe0xbebis/hyperpayalfredxw/nova
Stars141414
LanguageTypeScriptTypeScriptTypeScript
Setup difficultyeasyhardmoderate
Complexity2/55/53/5
Audiencedeveloperdeveloperwriter

Figures from each repo's GitHub metadata at analysis time.

How do you spin it up?

Difficulty · easy Time to first run · 5min
MIT license: free to use, modify, and distribute, including commercially, as long as the original copyright notice is kept.

Wtf does this do

Anscribe is a developer tool that connects terminal based user interfaces to AI coding assistants. It lets you press a keyboard shortcut inside a terminal app built with OpenTUI, a library for building rich terminal UIs, point at any visible element on screen, type an instruction, and immediately send that snapshot, called a Capture, to an AI agent. The Capture packages the UI element and your instruction as a formatted text payload. There are two delivery modes. The default clipboard mode copies the Capture to your system clipboard as markdown, which you can then paste into an AI assistant such as Claude Code, Cursor, or opencode. This mode works over SSH and inside dev containers with no extra setup. The optional MCP mode goes further: it stores Captures in a local SQLite database and runs a small local server that AI agents can connect to and pull from automatically, without manual pasting. Setup is minimal. A short code example in the README shows importing and calling a single function, passing a keybinding, and that is the entire integration. Press the shortcut, select a rendered element, add an instruction, press enter, and the Capture lands on the clipboard. The project is split across several packages: one for OpenTUI integration, one for MCP server functionality, one for React component enrichment, and a shared core package that all the others depend on. It is written in TypeScript, installed using the Bun package manager, and released under the MIT license.

Yoink these prompts

Prompt 1
Show me how to add anscribe's installCapture to my OpenTUI app with a custom keybinding.
Prompt 2
Explain the difference between anscribe's clipboard mode and its MCP mode.
Prompt 3
Help me set up the anscribe-mcp stdio server so my agent can pull Captures automatically.
Prompt 4
What does the @anscribe/core package do and why do the other packages depend on it?

Frequently asked questions

wtf is anscribe?

A tool that captures live terminal UI elements and sends them, with your instructions, to an AI coding assistant.

What language is anscribe written in?

Mainly TypeScript. The stack also includes TypeScript, Bun, OpenTUI.

What license does anscribe use?

MIT license: free to use, modify, and distribute, including commercially, as long as the original copyright notice is kept.

How hard is anscribe to set up?

Setup difficulty is rated easy, with roughly 5min to a first successful run.

Who is anscribe for?

Mainly developer.

View the repo → Decode another repo

This repo across BitVibe Labs

Don't trust strangers blindly. Verify against the repo.