filesystem-maintenance

amit-sw's avatarfrom amit-sw

Use when Codex needs to inspect or modify files via the local Python filesystem MCP server running at `servers/filesystem`.

⭐0stars🔀0forks📁View on GitHub🕐Updated Dec 24, 2025
[Files and Documents]

When & Why to Use This Skill

This Claude skill provides a robust interface for local filesystem management via a Python-based MCP server. It empowers AI agents to perform essential file operations—including directory listing, reading, writing, and deletion—within a secure, restricted environment. By bridging the gap between the AI and local storage, it facilitates complex tasks like automated refactoring, system maintenance, and real-time log analysis while maintaining strict security boundaries.

Use Cases

  • Automated Code Refactoring: Efficiently read source code and apply modifications across multiple files to streamline development workflows and ensure consistency.
  • Log Retrieval and Analysis: Access and inspect local system or application logs to diagnose errors, monitor performance, and troubleshoot issues without manual intervention.
  • Configuration Management: Safely modify configuration files, environment settings, and local metadata within a designated workspace root.
  • Workspace Discovery: Explore and map complex directory structures to provide the AI with necessary context for subsequent coding, auditing, or administrative tasks.
  • File Cleanup and Maintenance: Remove temporary artifacts, build logs, or obsolete files as part of a cleanup script or post-processing workflow.
namefilesystem-maintenance
descriptionUse when Codex needs to inspect or modify files via the local Python filesystem MCP server running at `servers/filesystem`.

Filesystem Maintenance

Purpose

Safely read, write, and delete files under MCP_FS_ROOT using the stdio-based filesystem MCP server bundled with this repository. Applies to refactors, log retrieval, and config updates when direct shell access is discouraged.

Setup Checklist

  1. Activate the repo virtual environment and install servers/filesystem/requirements.txt.
  2. Launch the server through mcp.json (python servers/filesystem/server.py) with MCP_FS_ROOT pointing at the workspace root.
  3. Confirm the server advertises the tools list_dir, read_file, write_file, and delete_path.

Workflow

  1. Discovery – call list_dir to understand the structure before editing. Limit breadth by passing the nearest parent directory.
  2. Read – fetch file contents using read_file(relative_path) to avoid stale local copies.
  3. Edit – stage modifications via normal repo tooling, then persist with write_file. When writing, pass relative paths (e.g., skills/github-operations/SKILL.md).
  4. Cleanup – remove temporary artifacts using delete_path only after verifying that the target path is inside MCP_FS_ROOT.

Notes

  • The _resolve guard inside server.py prevents escaping above MCP_FS_ROOT. If you see a permission error, double-check the path.
  • Use this skill when another skill (e.g., github-operations) references files you must read—chain the skills rather than reimplementing file access.
  • Keep writes idempotent: re-read the file to ensure committed changes match expectations.