Filesystem AI Agent
Give your AI agent scoped access to a local folder. Read, organize, and report on files with Oido — self-hostable, runs on your own machine, multi-provider LLM.
What you can automate with Filesystem + Oido
- Summarize every document dropped into a folder and post the digest to a channel.
- Search a directory of contracts for a clause without opening each file.
- Rename and sort incoming files by content on a schedule.
- Answer 'what's in this folder about X?' grounded in the actual files.
Oido connects to your local files through the Filesystem MCP server, which gives the agent read and write access to a directory you explicitly allow. Since Oido is self-hostable, this all stays on your own machine — useful when the data shouldn't leave it.
How it works
The Filesystem MCP server scopes the agent to the directories you pass as arguments — it can't read outside them. Oido runs the agent on demand or on a schedule. Combined with a channel, a local folder becomes something you can query in plain English from anywhere.
The allowed-path argument is the security boundary: list only the folders the agent should touch. Pair with Google Drive to bridge local and cloud files in one workflow.
Setup
Install the Filesystem MCP server as an Oido extension:
oido extensions install @modelcontextprotocol/server-filesystemOr add it directly to ~/.config/oido/mcp-servers.json:
{
"servers": {
"filesystem": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-filesystem",
"/path/to/allowed/dir"
],
"env": {},
"transport": "stdio"
}
}
}Confirm the exact source and env vars against the Filesystem MCP server — package names occasionally change.