pyautogui-mcp-server

Provides an MCP server (Streamable HTTP) that runs Python code in a fresh interpreter state with pyautogui instrumentation, returning captured stdout/stderr/results and inline screenshots (including mouse-operation previews). Includes a macOS-only helper to keep the display awake before automation.

Evaluated Apr 04, 2026 (0d ago)
Homepage ↗ Repo ↗ Automation mcp gui-agent pyautogui automation screenshots starlette uvicorn
⚙ Agent Friendliness
41
/ 100
Can an agent use this?
🔒 Security
28
/ 100
Is it safe for agents?
⚡ Reliability
20
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
55
Documentation
65
Error Messages
0
Auth Simplicity
10
Rate Limits
5

🔒 Security

TLS Enforcement
20
Auth Strength
20
Scope Granularity
0
Dep. Hygiene
60
Secret Handling
50

No authentication/authorization guidance is provided for the MCP server in the README/manifest. The tool executes Python in a fresh interpreter with pyautogui instrumentation—this implies the ability to control the local UI and capture screenshots; without sandboxing and request authorization, this is high-risk if exposed to untrusted clients. TLS enforcement and secret-handling behavior are not documented; treat deployment as internal/trusted-only.

⚡ Reliability

Uptime/SLA
0
Version Stability
30
Breaking Changes
30
Error Recovery
20
AF Security Reliability

Best When

You control the client/agent and want an MCP tool to execute pyautogui-driven UI steps and receive logs + screenshots as structured tool outputs.

Avoid When

You need strong security boundaries (e.g., running untrusted code) or you cannot grant the server/API full control over UI automation and screen capture.

Use Cases

  • GUI automation via an MCP-capable agent
  • Tool-based execution of Python that drives the mouse/keyboard with pyautogui
  • Remote/agent-controlled screenshot capture and inline delivery
  • Workflow automation on macOS where the display must stay awake

Not For

  • Production-safe execution of untrusted code (no sandboxing guarantees are described)
  • Highly regulated environments requiring strict audit/compliance controls
  • Environments where controlling the OS UI is disallowed

Interface

REST API
No
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
No
Webhooks
No

Authentication

OAuth: No Scopes: No

The README and manifest do not describe authentication/authorization for the MCP server. The server is started with --host/--port only.

Pricing

Free tier: No
Requires CC: No

Open-source library (MIT) installed via pip; no hosted pricing described.

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • GUI automation is inherently stateful; retries can repeat mouse actions unless the agent adds its own guardrails
  • Running Python code remotely can be dangerous unless you fully trust the caller; no sandboxing guarantees are described
  • Screenshot coordinates may drift if display scaling/coordinate mapping differs; normalization is mentioned but not configured/documented in detail

Alternatives

Full Evaluation Report

Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for pyautogui-mcp-server.

AI-powered analysis · PDF + markdown · Delivered within 30 minutes

$99

Package Brief

Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.

Delivered within 10 minutes

$3

Score Monitoring

Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.

Continuous monitoring

$3/mo

Scores are editorial opinions as of 2026-04-04.

8642
Packages Evaluated
17761
Need Evaluation
586
Need Re-evaluation
Community Powered