March 18, 2026
Everything’s a file… and a fight
Aggregated File System (AGFS), a modern tribute to the spirit of Plan 9
Dev world split: genius shortcut or Rube Goldberg for bots
TLDR: AGFS turns cloud services into files so bots and humans can use “cat” and “echo” instead of many APIs. Commenters are split between nostalgia-fueled curiosity and eye-rolls, with skeptics calling it a Rube Goldberg stunt for LLMs and pedants arguing it’s not real Plan 9, just REST in cosplay.
“Everything is a file” is back—with AI drama. The new AGFS project turns cloud services (like queues and databases) into simple files you can “cat,” “echo,” and “ls.” It even mounts like a real folder on Linux via FUSE, aiming to let humans and AI bots use one easy interface. Think: copy a file to send to storage, read a file to get a message. Simple—at least on paper.
The crowd? Spicy. One commenter sneered, “More LLM focused crap,” framing AGFS as yet another AI gimmick. Another labeled it a “Rube Goldberg contraption,” asking for proof it beats just giving an AI a Python prompt. Meanwhile, the nostalgia squad swooned over Plan 9 vibes, but wondered if this is just files on top of the same old web calls. As one put it: why copy a file to S3 when you’re still… calling S3?
Then came the pedant brigade: “This isn’t Plan 9’s 9p protocol—it’s REST,” sparking a mini flame-war over purity vs practicality. Jokes flew: “I just ‘cat’d my database’,” “mkdir my feelings,” and our favorite: “Everything’s a file? Everything’s a fight.”
If it works, AI tools and scripts could share one brain-dead-simple way to talk to everything. If it doesn’t, it’s just cosplay for sysadmins and bots—cute pipes, no flow.
Key Points
- •AGFS maps diverse backend services (queues, KV stores, object storage, SQL) to filesystem operations, inspired by Plan 9.
- •It offers both an HTTP/REST interface and optional Linux FUSE mounting, enabling standard file reads/writes by any program.
- •Installation is available via curl script or Docker; FUSE mounting in Docker is not supported on macOS.
- •Examples show usage for key-value storage, message queuing, SQL querying via session files, and agent heartbeat tracking.
- •Benefits cited include a unified interface, LLM-friendly commands, shell-based composability, and easy debugging using ls and cat.