hands-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@hands-mcpWhat's the savings balance for member 10041?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
hands-mcp
An MCP server for hands.
hands records a UI flow in an old web app once and replays it without a model. This server reads the recorded capabilities at startup and registers each one as an MCP tool. An agent can then call legacycu__member_savings_balance with a member number and get the balance back.
Built with mcp-use.
Run
You need Node 22 and a working checkout of hands. Its README covers the Python setup.
From the hands checkout, start the demo app that the capabilities run against:
python -m hands.cli serve-target --port 8080Then in this folder:
npm install
npm run devHANDS_DIR is the path to the hands checkout (default ../computer-use-automation). HANDS_PYTHON is the interpreter (default python). Point it at the venv's python if the venv is not active.
Open http://localhost:3000/mcp/inspector and call a tool. Member 10041 has a savings balance of 15.02. Member 99999 does not exist. That run returns a MEMBER_NOT_FOUND outcome and isError stays false.
Related MCP server: nuphus-mcp
How it works
hands catalog prints every capability as a tool definition with a JSON Schema for its inputs. The server turns each schema into a zod schema, so a bad member number is rejected before any browser starts.
A tool call runs hands invoke and returns its JSON result as structured content. A run that ends in failure sets isError.
Runs have no operator and no grant for steps that change data. legacycu__open_subaccount therefore stops at the hands policy and returns a POLICY_BLOCKED failure.
Each run writes its screenshots and step log under evidence/ in this folder. The result carries the path.
This server cannot be deployed
Maintenance
Related MCP Connectors
Discover and call AI agents via MCP. Supports A2A agents and platform agents with async tasks.
Human-input bridge for AI agents with voice-first answer links, MCP tools, and HTTP APIs.
Pay-per-use tool marketplace for AI agents. Search, price-check, and call APIs via MCP.
Let AI agents query data and act across all your business apps via MCP.
Related MCP Servers
- AlicenseAqualityFmaintenanceProvides 14 MCP tools for AI agent infrastructure, enabling knowledge base queries, skill search, handoffs, blueprint validation, trust scoring, identity verification, SLA validation, and compliance checks.22MIT
- AlicenseNot gradedqualityBmaintenanceProvides desktop automation capabilities (e.g., click, browser snapshot) as MCP tools for DSH agents, enabling computer use via natural language.105 npm4MIT

astrolabe-mcpofficial
AlicenseNot gradedqualityBmaintenanceEnables agents to call a website's own authenticated GraphQL and REST APIs directly as typed MCP tools, derived from captured browser sessions instead of scraping HTML.MIT- AlicenseNot gradedqualityAmaintenanceEnables capturing successful agent tool calls into reviewed, parameterized routines that can be replayed with different inputs across MCP, DSH, and WebMCP workflows.123 npm1MIT