Show HN: A durable filesystem layer for AI agents
github.comI run AI agents on my laptop and cloud. Often I wish to synchronize the memory markdowns created on multiple platforms. So I built a S3 based durable filesystem which can be mounted anywhere.
It is implemented in Rust with SDK in both Python, TypeScript and a CLI for agents.
How are you thinking about concurrent mounts and conflict handling if two agent runtimes write to the same SmolFS volume?
SmolFS has a distributed posix filesystem. we use juicefs which separates the actual data and metadata (permissions, file size changes etc). Metadata store provides atomicity and locking primitives which keeps the file system horizontally scalable while maintaining concurrent mounts and handle concurrent agent operations.
ooo, I was looking for something like this when I was designing our agent architecture. Wish this existed before. Ended up building something quite similar, although this looks more complete and modular. Nice!
Nice to know that others also faced the same issue. Did you try something like archil or agentfs from turso?
Interesting. Good use case especially for portability to cloud.
Yeah, it’s quite handy that I can mount the same storage on my macbook and then somewhere on a random cloud.