AX Local Operations

You can turn any MCP-enabled chat application into a capable local assistant by running AX Local Operations MCP Server. It provides secure local file handling, line editing, searching, comparing, hashing, permissions, archiving, monitoring, command execution, and task management, enabling practical, real-world interactions with your computer through AI.

How to use

To use AX Local Operations with an MCP client, configure the client to connect to the local MCP server and ensure the server is running in the background. The AI application will first discover available tools, then issue requests to perform tasks such as reading or writing files, editing content line by line, running commands in a safe sandbox, or managing long-running tasks. You can leverage the environment memory to persist context across sessions and keep your working state organized.

Typical workflows involve starting the MCP server, connecting your MCP client, and then issuing instructions that map to the provided tools. For instance, you can read a file, modify specific lines, search code bases with patterns, compare two files to see differences, or create tasks with due dates and subtasks. The system enforces safety checks and path restrictions to minimize risk while enabling productive automation.

How to install

Prerequisites: Node.js version 18.0.0 or newer, and npm or yarn. Ensure your environment can run JavaScript tools locally.

Install the MCP server globally so you can access the executable from any client. Run the following command in your terminal:

npm install -g ax-local-operations-mcp

Once installed, you can start the MCP server directly or invoke it via npx in clients that support MCP configurations.

npx -y ax-local-operations-mcp

Additional sections

Configuration and usage details shown here describe how to connect with popular MCP clients such as LM Studio and Qwen, including the exact commands and environment settings used to run the local server.

Configuration and safety notes

The server uses a layered safety model to control file access and command execution. Path sandboxing restricts access to the user’s home directory by resolving relative paths to the user’s home. Commands are filtered with a policy that can deny or require confirmation for dangerous operations. Errors are standardized and include clear codes like E_PATH_DENIED and E_DANGEROUS_CMD. Recursive depth limits and safe spawning practices are enforced for security.

Troubleshooting

If you encounter connection issues, verify Node.js paths and file permissions. Ensure the MCP server is running and the client is configured to connect via the correct stdio interface. Check logs for errors and confirm the configuration blocks are valid JSON within the client’s MCP integration.

Notes on environments and security

Environment memory stores session-scoped data in a dedicated file, enabling consistent context usage across interactions. Security checks apply to all paths and commands, and the server will return structured error responses if you attempt unsafe operations.

Available tools

file_operation

Read, write, list, create, and delete files and directories.

file_edit

Line-level editing including delete_lines, insert_lines, replace_lines, and append_lines.

file_search

Search within files with patterns, including regex and file type filters.

file_compare

Diff two files and show line-level differences.

file_hash

Compute hashes with MD5, SHA1, SHA256, SHA512.

file_permissions

Modify and display file permissions with recursive options.

file_archive

Compress and extract archives in multiple formats.

file_watch

Monitor file changes and events in real time.

execute_command

Run local system commands with safety protections.

task_manager

Create, list, update, complete, and organize tasks.

time_tool

Get current time, timestamps, and locale times with zone support.

environment_memory

Store and retrieve environment-related information across sessions.