VERIFIED

Safe Local Python Executor

Safety: 65/100

0.0(0)

Executes Python code snippets locally within a restricted environment, offering a safer alternative to direct evaluation for LLM-generated code.

MCP ServersFreemium

Boost this tool

Subscribe to listing upgrades or segmented pushes.

Overview

Executes Python code snippets locally within a restricted environment, offering a safer alternative to direct evaluation for LLM-generated code.

This executor provides a balance between convenience and security. It's safer than direct Python execution but less secure than containerized solutions. Use with caution, especially with untrusted code sources.

Performance is limited by the LocalPythonExecutor's overhead and the restricted environment. Complex or long-running code may be slow.

No direct cost, but excessive use can consume local resources.

HybridPython44 starsMIT

Connections & Capabilities

Connects To

GitHubDocker

Capabilities

Exec

Client Compatibility

Claude Desktop

full

Cursor

untested

Windsurf

untested

VS Code (Copilot)

untested

Quickstart

Install

npx -y @smithery/cli install @maxim-saplin/mcp_safe_local_python_executor --client claude

Claude Desktop Config

{
    "mcpServers": {
        "safe-local-python-executor": {
            "command": "uv",
            "args": [
                "--directory", 
                "/path/to/mcp_local_python_executor/",
                "run",
                "mcp_server.py"
            ]
        }
    }
}

Tools (1)

moderate

run_python

Executes a Python code snippet and returns the output.

Executes arbitrary code, even within a sandbox.

Security

Auth Method

None

Data Locality

hybrid

Known Risks

!Potential for code injection if the LLM generates malicious code.
!Resource exhaustion if the executed code enters an infinite loop.
!Bypasses in the LocalPythonExecutor's security measures.
!Information leakage through allowed imports (e.g., timing attacks).

Safety Score

moderate risk

Positive

+Uses Hugging Face's LocalPythonExecutor for sandboxing.
+Restricted list of allowed Python imports.
+No file I/O operations permitted.
+Runs in a Python venv via uv.

Risks

-Code execution, even sandboxed, carries inherent risks.
-Vulnerable to bypasses in the LocalPythonExecutor.
-Potential for resource exhaustion if code isn't well-behaved.
-Limited isolation compared to Docker or VMs.

Use Cases

Mathematical Calculations

-Calculating complex equations
-Performing statistical analysis
-Generating mathematical sequences

Data Manipulation

-Parsing and processing text data
-Converting data formats
-Filtering and sorting data

Algorithmic Tasks

-Implementing sorting algorithms
-Searching for patterns in data
-Solving logic puzzles

Simple Simulations

-Simulating coin flips or dice rolls
-Modeling basic physical systems
-Generating random data for testing

Who Is This For?

Best For

Prototyping Python-based tools
Executing simple calculations
Performing data transformations
Educational purposes

Not Ideal For

Production environments requiring high security
Tasks requiring file I/O or external network access
Long-running or resource-intensive computations
Applications handling sensitive data

Autonomy & Execution

Default ModeConfigurable

Sandboxed

The level of autonomy depends on how the 'run_python' tool is integrated into the agent's workflow and how prompts are constructed.

StatelessYes

Retry LogicNot documented

ConcurrencySingle-threaded

Production Tip

Implement timeouts and resource limits to prevent runaway code from consuming excessive resources.

FAQ

What are the limitations of the allowed Python imports?

Only a restricted set of standard libraries are available, preventing access to potentially dangerous functions.

How secure is the LocalPythonExecutor?

It provides a degree of sandboxing, but it's not a substitute for full isolation like a VM or container.

Can I use this to execute arbitrary Python code from the internet?

While possible, it's strongly discouraged due to the inherent risks of executing untrusted code.

What happens if the code I execute has an infinite loop?

The execution will continue until manually stopped or a resource limit is reached, potentially freezing the application.

Does this support multi-threading?

No, the execution environment is single-threaded.

Can I install additional Python packages?

No, only the pre-approved standard libraries are available.

How do I configure the timeout for code execution?

The timeout is not configurable. You may need to implement your own timeout logic within the code.

Metrics

Discovered2/14/2026

Reviews0

Saved By0 users

Related Tools

Safe Local Python Executor - AI Tool Review & Alternatives | Flaex AI