# Node Mantis [Health: Active]

**Category:** 📂 Browser Automation  
**Repository:** https://nodemantis.com/docs/mcp  
**npm Downloads (last month):** 1692  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/node-mantis

## Description
Lets AI agents complete tasks on websites: give a goal in plain English and it clicks, types and fills in forms like a person would, then returns data matching your JSON Schema.

## Claude Desktop Quick Installation
Install path inferred — verify against the README before running it. Uses `npx` (confidence: medium):

```json
"mcpServers": {
  "node-mantis": {
    "command": "npx",
    "args": ["-y","node-mantis","mcp"]
  }
}
```

## Documentation

## What Node Mantis does

Node Mantis gives AI agents a way to complete tasks on websites through MCP. An agent can provide a goal in plain English instead of specifying every browser action individually. Node Mantis then carries out the interaction by using actions such as clicking, typing, and filling in forms.

The server is suited to workflows where the agent must both change information on a website and collect a result. For example, a task can involve entering values into a web form and returning selected data after the interaction finishes. The output follows a JSON Schema supplied with the task, giving the calling agent a defined structure for the returned information.

## How it works

A typical workflow begins with a natural-language instruction describing the website task. Node Mantis interprets that goal and performs the required page interactions. Its browser automation behavior is based on common user actions: selecting controls, entering text, and submitting or completing forms.

After the website interaction, the agent receives data shaped according to the requested JSON Schema. This makes the result suitable for workflows that need predictable fields rather than an unstructured page description. The schema also lets the calling workflow define the expected form of the extracted data.

The main abstraction is task completion rather than individual low-level browser commands. Agents describe the outcome they need, while Node Mantis handles the corresponding sequence of clicks, text entry, and form completion on the website.

## Tools and capabilities

- Accepts website tasks written as plain-English goals.
- Clicks controls as part of a website workflow.
- Types text into website interfaces.
- Fills out web forms.
- Returns data matching a supplied JSON Schema.
- Supports workflows that combine website interaction with structured data collection.

Node Mantis belongs in the browser automation category because its scope is interaction with websites. It is most relevant when an agent needs to follow a user-like sequence of actions instead of only retrieving already-structured information through an API. The JSON Schema output is useful when downstream code needs consistent fields from each completed task.

The server’s value depends on describing the desired website outcome clearly and defining the data shape the agent should return. A goal can express the task in human terms, while the schema establishes the expected structure of the result.

_Full upstream README: https://allmcps.com/mcp/node-mantis/readme_

