Skip to main content
Glama
tjariy-code-interpreter-mcp-server

tjariy-code-interpreter-mcp-server

by jariy17

tjariy-code-interpreter-mcp-server

An MCP (Model Context Protocol) server that provides secure code execution capabilities using AWS Bedrock AgentCore's CodeInterpreter.

Features

  • Execute Python, JavaScript, and TypeScript code in a secure sandbox

  • Run shell commands

  • File operations (read, write, list, remove)

  • Session management (auto-managed by default)

Related MCP server: e2b-mcp-server

Prerequisites

  • Node.js 20+

  • AWS credentials configured (via environment variables, AWS profile, or IAM role)

  • Access to AWS Bedrock AgentCore CodeInterpreter service

Installation

npm install
npm run build

Configuration

Set the AWS region via environment variable (default: us-west-2):

export AWS_REGION=us-west-2

AWS credentials can be configured via:

  • Environment variables (AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY)

  • AWS credentials file (~/.aws/credentials)

  • IAM role (when running on AWS infrastructure)

Usage

Running the Server

# Production
npm start

# Development
npm run dev

Claude Desktop Configuration

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "code-interpreter": {
      "command": "npx",
      "args": ["-y", "tjariy-code-interpreter-mcp-server"],
      "env": {
        // Optional: defaults to "us-west-2"
        "AWS_REGION": "us-west-2",
        // Optional: defaults to "default" profile, or uses IAM role if on AWS
        "AWS_PROFILE": "your-profile"
      }
    }
  }
}

Available Tools

Tool

Description

execute_code

Execute Python/JavaScript/TypeScript code in sandbox

execute_command

Run shell commands

read_files

Read file contents from sandbox

write_files

Write files to sandbox

list_files

List directory contents

remove_files

Delete files

start_session

Create a new interpreter session

stop_session

Terminate current session

get_session

Get session information

list_sessions

List all sessions

Examples

Once configured with Claude Desktop, you can ask Claude to:

  • "Run this Python code: print('Hello, World!')"

  • "Create a file called test.py with a fibonacci function"

  • "List all files in the current directory"

  • "Execute a shell command to check the Python version"

License

MIT

Available Tools

10 tools
execute_codeC

Execute code in a secure sandboxed environment. Supports Python, JavaScript, and TypeScript.

Inputs: code (required), language.

execute_commandC

Execute a shell command in the sandbox environment

Inputs: command (required).

get_sessionB

Get information about a specific session

Inputs: sessionId (required).

list_filesA

List files in a directory in the sandbox

Inputs: path.

list_sessionsA

List all code interpreter sessions

Inputs: maxResults.

read_filesA

Read file contents from the sandbox filesystem

Inputs: paths (required).

remove_filesB

Remove files from the sandbox filesystem

Inputs: paths (required).

start_sessionC

Start a new code interpreter session

Inputs: name, timeout.

stop_sessionA

Stop the current code interpreter session

Inputs: no named parameters.

write_filesC

Write files to the sandbox filesystem

Inputs: files (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.6/5.0

    Scored across 10 tools

    Disambiguation4/5

    execute_code and execute_command both involve running something, which could cause minor confusion, but the former is for Python/JavaScript/TypeScript code and the latter is for shell commands. File operations and session operations are each clearly distinct from one another.

    Naming Consistency5/5

    All tools follow a consistent verb_noun snake_case pattern such as execute_code, list_files, start_session, and stop_session. The naming is predictable and easy to navigate.

    Tool Count5/5

    10 tools is well-scoped for a code interpreter sandbox, covering execution, shell commands, file management, and session lifecycle without redundancy. Each tool serves a clear and necessary purpose.

    Completeness5/5

    The toolset provides complete lifecycle coverage for sessions and files, plus code and shell execution. There are no obvious missing operations that would prevent an agent from accomplishing typical sandbox tasks.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      D
      maintenance
      A general-purpose MCP server providing web search, persistent memory storage, and secure code execution capabilities. It enables AI agents to search the web, store and retrieve data, and run Python/JavaScript code in sandboxed environments.
      8
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Production-grade MCP server that enables AI assistants to execute code securely in isolated E2B sandboxes.
      Apache 2.0
    • A
      license
      Not graded
      quality
      D
      maintenance
      A self-hosted MCP server providing stateful code execution (Python/JavaScript) with SQL query support, file I/O, and artifact management, replacing Azure Code Interpreter.
      MIT
    • F
      license
      Not graded
      quality
      D
      maintenance
      An MCP server that executes TypeScript/JavaScript code in a sandboxed Bun environment with permission-based security controls, code snippets, and a web management UI.
      -