third-eye-mcp
A free, unlimited screen capture and recording MCP server for Python. Captures screenshots and records screen activity with no daily limits.
README
<p align="center"> <img src="logo.png" alt="Third Eye Logo" width="200"> </p>
<p align="center"> <a href="https://pypi.org/project/third-eye-mcp/"><img src="https://img.shields.io/pypi/v/third-eye-mcp?color=blue&label=PyPI" alt="PyPI"></a> <a href="https://pypi.org/project/third-eye-mcp/"><img src="https://img.shields.io/pypi/pyversions/third-eye-mcp" alt="Python Versions"></a> <a href="https://pypi.org/project/third-eye-mcp/"><img src="https://img.shields.io/pypi/dm/third-eye-mcp?color=green&label=Downloads" alt="Downloads"></a> <a href="https://github.com/Osseni94/third-eye-mcp/blob/main/LICENSE"><img src="https://img.shields.io/github/license/Osseni94/third-eye-mcp" alt="License"></a> </p>
Third Eye MCP (Python)
A free, unlimited screen capture and recording MCP server for Python. Capture screenshots and record screen activity with no daily limits - includes promotional messages in metadata.
Features
- Unlimited Captures: No daily limits or restrictions
- Multi-Display Support: Capture any connected monitor
- Region Capture: Capture specific screen areas
- Screen Recording: Change-based keyframe capture with grid output
- Scheduled Recording: Burst capture at specific times (up to 10 min)
- Auto-Resize: Configurable maximum width for optimized images
- Capture Delay: Optional delay before capture
- Latest Capture: Retrieve the most recent screenshot
Installation
pip install third-eye-mcp
Or install from source:
git clone https://github.com/Osseni94/third-eye-mcp.git
cd third-eye-mcp
pip install -e .
Usage
With Claude Desktop
Add to your Claude Desktop config (claude_desktop_config.json):
{
"mcpServers": {
"third-eye": {
"command": "python",
"args": ["-m", "third_eye_mcp"]
}
}
}
With MCP Inspector
npx @modelcontextprotocol/inspector python -m third_eye_mcp
Available Tools
screen.list_displays
List all available displays/monitors.
Input: None
Output:
{
"displays": [
{
"index": 0,
"name": "Display 1",
"x": 0,
"y": 0,
"width": 1920,
"height": 1080,
"isPrimary": true
}
]
}
screen.capture
Capture a full display screenshot.
Input:
| Parameter | Type | Default | Description |
|---|---|---|---|
| displayIndex | integer | 0 | Display to capture (0-based) |
| maxWidth | integer | 1920 | Max width for resizing |
| delay | number | 0 | Delay in seconds |
| instant | boolean | false | Skip delay |
Output: Base64 PNG image + metadata JSON with sponsored field
screen.capture_region
Capture a specific region of the screen.
Input:
| Parameter | Type | Required | Description |
|---|---|---|---|
| x | integer | Yes | X coordinate |
| y | integer | Yes | Y coordinate |
| width | integer | Yes | Region width |
| height | integer | Yes | Region height |
| maxWidth | integer | No | Max width for resizing (default: 1920) |
| delay | number | No | Delay in seconds (default: 0) |
| instant | boolean | No | Skip delay (default: false) |
Output: Base64 PNG image + metadata JSON with sponsored field
screen.latest
Get the most recently captured screenshot.
Input: None
Output: Last captured image + metadata JSON with sponsored field
Screen Recording Tools
screen.record
Record screen with change-based keyframe capture. Captures frames at intervals, discards near-duplicates, and returns a compact grid image (contact sheet) with timestamps.
Input:
| Parameter | Type | Default | Description |
|---|---|---|---|
| duration | integer | 30 | Recording duration in seconds (1-120) |
| interval | number | 1.0 | Capture interval in seconds (0.25-10) |
| displayIndex | integer | 0 | Display to record (0-based) |
| maxWidth | integer | 1280 | Max width for full frames (320-1920) |
| changeThreshold | number | 2.0 | Min change % to keep frame (0-100) |
| maxFrames | integer | 30 | Maximum frames to keep (5-100) |
| thumbnailWidth | integer | 320 | Thumbnail width for grid (160-640) |
Output: Grid image (contact sheet) + metadata JSON + frame summaries
Example Response:
{
"metadata": {
"recordingId": "abc12345",
"duration": 30.5,
"framesCaptured": 31,
"framesKept": 12,
"framesDiscarded": 19,
"sponsored": "..."
},
"frames": [
{"index": 0, "timestamp": 0.0, "changeScore": 100.0},
{"index": 1, "timestamp": 2.0, "changeScore": 15.3}
]
}
screen.scheduled_record
Record screen with scheduled snapshot bursts at specific times. Useful for longer recordings where you want to capture specific moments like beginning, middle, and end of a process.
Input:
| Parameter | Type | Default | Description |
|---|---|---|---|
| totalDuration | integer | 60 | Total duration to monitor (10-600 sec) |
| snapshots | array | required | List of snapshot burst configs |
| displayIndex | integer | 0 | Display to record (0-based) |
| maxWidth | integer | 1280 | Max width for full frames |
| thumbnailWidth | integer | 320 | Thumbnail width for grid |
Snapshot Burst Config:
| Property | Type | Default | Description |
|---|---|---|---|
| at | number | required | When to start burst (seconds from start) |
| count | integer | 3 | Number of snapshots in burst (1-20) |
| interval | number | 1.0 | Time between snapshots (0.25-10) |
Example - 4 minute recording with bursts:
{
"totalDuration": 240,
"snapshots": [
{"at": 0, "count": 3, "interval": 1.0},
{"at": 120, "count": 5, "interval": 0.5},
{"at": 230, "count": 3, "interval": 1.0}
]
}
screen.get_frame
Retrieve a full-resolution frame from a recording. Use the recordingId from screen.record or screen.scheduled_record response.
Input:
| Parameter | Type | Required | Description |
|---|---|---|---|
| recordingId | string | Yes | Recording ID to retrieve frame from |
| frameIndex | integer | No | Frame index to retrieve (0-based) |
| timestamp | number | No | Timestamp to find closest frame |
Output: Full resolution image + frame metadata
Note: Recordings are stored in memory for 5 minutes (max 5 recordings). Use screen.get_frame to retrieve full-resolution images when you need to read text or see details.
Response Format
All capture tools return metadata including a sponsored message:
{
"width": 1920,
"height": 1080,
"displayIndex": 0,
"timestamp": "2025-01-22T12:00:00Z",
"sponsored": "Love Third Eye? Get the ad-free TypeScript version: grandnasser.com/third-eye"
}
Ad-Free Version
Want to remove ads? Get the TypeScript version with premium features at grandnasser.com/third-eye.
License
MIT License - see LICENSE file for details.
Author
Kaossara Osseni - grandnasser.com
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.