Manage files across local and remote systems with encoding-aware, indentation-smart tools. Built for AI agents that need SSH, SFTP, and 20+ file operations without breaking code formatting or character encoding.
io.github.knewstimek/agent-tool
Local install
STDIO
No auth required
How models use it and what it is built for.
Manage files across local and remote systems with encoding-aware, indentation-smart tools. Built for AI agents that need SSH, SFTP, and 20+ file operations without breaking code formatting or character encoding.
Local install — runs as a subprocess.
Where to find authoritative docs and source for agent-tool.
Paste any of these into Agent Studio after connecting agent-tool.
Common questions about connecting and running agent-tool.
What file operations does agent-tool support?
The server provides 20+ tools for file management including read, write, create, delete, and list operations. It handles SSH and SFTP connections for remote file access, with built-in awareness of encoding and indentation to prevent corruption of code files.
How do I set up SSH or SFTP access?
The server supports SSH and SFTP transport. Refer to the GitHub repository (https://github.com/knewstimek/agent-tool) for configuration details on setting up remote credentials and connection parameters.
Does it preserve file formatting when editing code?
Yes. The tool is indentation-smart and encoding-aware, meaning it detects and preserves your file's original indentation style (spaces vs. tabs) and character encoding (UTF-8, etc.) when reading and writing.
Can I use this with both local and remote files?
Yes. The server supports local file operations as well as remote access via SSH and SFTP, allowing you to work with files across different systems in a single agent workflow.
What happens if a file has mixed line endings or unusual encoding?
The encoding-aware design detects and handles various character encodings and line ending styles (CRLF, LF). This prevents common issues where file operations corrupt binary data or break cross-platform compatibility.
MCP Playground runs 10,000+ hosted MCP servers — GitHub, Linear, Notion, Stripe, Sentry and more — across Claude, GPT, Gemini, DeepSeek and 30+ AI models. Compare model answers side-by-side, save agent presets, share runs. Zero install.
Open Agent Studio