v2.8.2

Agents are good at working with files, but the filesystem in most setups is a scratch directory that vanishes when the run ends. Fine for a one-shot task, useless for anything an agent is supposed to remember. Agno's new FileSystem is a durable text store an agent writes to and reads back across runs, so the decision it recorded and the checkpoint it saved are still there when it starts up again in a fresh process.

from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.fs import FileSystem

fs = FileSystem(SqliteDb(db_file="tmp/filesystem.db"))

agent = Agent(
    tools=[fs.tools()],
    instructions=["You keep durable working notes.", fs.instructions()],
)

agent.print_response("Record in notes/decisions.md: SQLite for dev, Postgres for prod.")
# run again in a new process and the note is still there to read back

‍

You decide where the files actually live. Point it at SQLite for local development, Postgres for a deployed app running multiple workers, or local disk when you want to open the files in an editor yourself. The agent code doesn't change when you switch.

Namespaces are what make it safe to point at real users. Give it a templated namespace like "assistant/{user_id}" and each user's files are scoped to their own space, resolved from the run context at call time. The model can't reach into another namespace by passing a different path in a tool call, and if the user id is missing, the operation is blocked instead of running against the wrong space. You still enforce authorization at the backend, but the agent itself can't wander out of the namespace you put it in.

View the FileSystem docs to learn more.