DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to let an AI agent drive a running app through MCP with Limoni

Connect Claude Code or Cursor to a running Limoni app so the agent can click, type and read widgets.

Authorization notice. Limoni is dual-use. Only use it on software, sites and systems you own or have explicit permission to test.

Credentials. The agent can type into fields such as a deploy token, so only run this against a debug build with test credentials.

Who it is forA developer who wants an AI coding agent to walk through a terminal workflow, such as a release checklist, and report results.

The problemAgents cannot reliably use a terminal UI that only exposes raw characters on screen.

The outcomeAn agent that reads the semantic tree, clicks buttons, types into fields and waits for text, using eight MCP tools, while secret fields stay hidden from it.

What you need first

  • Go toolchain installed (for go install, go get, go run and go test)
  • A terminal that supports the app's output (images and some features depend on the terminal)
  • For the MCP use case: an MCP client such as Claude Code or Cursor, and limoni-mcp available on PATH (the README does not show how to install it)
  • For the backdrop use case: curl, sh, and a terminal emulator

Steps

  1. Run go run -tags limoni_debug ./examples/agent_checklist in one terminal and leave it running
  2. Run claude mcp add limoni -- limoni-mcp -socket "$XDG_RUNTIME_DIR/limoni-checklist.sock" to register the server with Claude Code
  3. Ask the agent to read the tree, then work through the checklist using the click, type_text and wait_for tools
  4. Check the checklist state in the running app after the agent finishes

Prompt for your agent

Using the limoni MCP server in this repo at <path>, read the semantic tree of the running agent_checklist app, complete the checklist items named <items>, and report the final tree. Verify that no secret field value is returned in your output.

Caveat. The automation socket exists only in -tags limoni_debug builds and is closed by default, so the command above does not work in a normal build.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for thebanri/limoni and "Let an AI agent drive a running app through MCP"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer