mcp-run-python

mcp-run-python is an MCP server that executes Python code inside a sandboxed WebAssembly environment by running Pyodide in Deno. It supports running async code, capturing stdout/stderr/return values, and provides helpers for preparing the Deno runtime/environment and for running code via an MCP stdio transport or streamable HTTP transports.

Evaluated Mar 30, 2026 (21d ago)
Repo ↗ Ai Ml mcp python sandboxing pyodide deno code-execution tooling
⚙ Agent Friendliness
60
/ 100
Can an agent use this?
🔒 Security
34
/ 100
Is it safe for agents?
⚡ Reliability
31
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
75
Documentation
70
Error Messages
--
Auth Simplicity
95
Rate Limits
0

🔒 Security

TLS Enforcement
50
Auth Strength
10
Scope Granularity
0
Dep. Hygiene
60
Secret Handling
60

The README indicates code runs in Pyodide within Deno, with isolation from the OS, but explicitly warns that Pyodide can run arbitrary JavaScript (tainting runtime across invocations, reading/writing files accessible to the runtime, and OOM/resource exhaustion). No authentication is documented, and network safety (auth, rate limiting, timeouts) is not described, so deployment hardening is required. Dependency hygiene is unknown from the excerpt; only mcp dependency is visible in the manifest.

⚡ Reliability

Uptime/SLA
0
Version Stability
35
Breaking Changes
40
Error Recovery
50
AF Security Reliability

Best When

Used locally or in a tightly controlled environment where you can control who can submit code and you understand the sandbox limitations; best for short-lived tool calls from trusted agent clients.

Avoid When

Avoid for public HTTP execution of untrusted code, multi-tenant SaaS execution, or situations requiring rigorous security guarantees stronger than the documented sandbox caveats (pyodide can run arbitrary JS in the same runtime).

Use Cases

  • Running LLM-generated Python snippets in an isolated(ish) sandbox for analysis or computation
  • Integrating Python execution as a tool in an MCP-based agent workflow (e.g., via Pydantic AI MCPServerStdio)
  • Automated evaluation or data-processing tasks where user-provided code needs some isolation
  • Lightweight interactive execution of multiple code blocks using a persistent sandbox session

Not For

  • Running untrusted/malicious code without strict operational safeguards and resource limits
  • Any workload that requires strong isolation from JavaScript/runtime-level side effects
  • Production deployments that need a stable, supported service (the project is archived/retired)
  • Exposing a network-facing execution endpoint to the public without authentication and rate limiting

Interface

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

Authentication

OAuth: No Scopes: No

README does not describe any authentication mechanism. If using streamable-http transports in a networked setting, auth/rate limiting would need to be provided externally or by transport-layer/security measures not documented here.

Pricing

Free tier: No
Requires CC: No

Self-hosted open-source tooling; no vendor pricing described.

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • Project is archived/retired; reliability and security posture may degrade over time.
  • Security caveat: Python running in Pyodide can execute arbitrary JavaScript in the runtime, with potential for runtime tainting and resource exhaustion if not controlled.
  • Dependency installation is a two-step Deno process; first-run behavior may incur latency and should be accounted for in agent timeouts.
  • No explicit auth/rate-limit behavior is documented; network-facing usage requires extra safeguards outside the package.

Alternatives

Full Evaluation Report

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

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 2026-03-30.

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