Skip to content

Agno

Add Suprflo long-term memory to Agno agents. Ships in the Python SDK.

Install

bash
pip install "suprflo[agno]" "agno<2"

Targets agno 1.x. Agno 2.x replaced the agno.memory.db.base.MemoryDb contract this adapter implements, so importing the module under 2.x raises an ImportError that says so.

What you get

Both are built on the Python SDK MemoryClient:

  • SuprfloTools — an Agno Toolkit registering save_memory and search_memory, bound to one client and user_id.
  • SuprfloMemoryDb — a MemoryDb that stores Agno's agent memories in Suprflo instead of a local table.

Tools for an agent

python
from agno.agent import Agent
from suprflo import MemoryClient
from suprflo.integrations.agno import SuprfloTools

client = MemoryClient(api_key="YOUR_API_KEY")
agent = Agent(tools=[SuprfloTools(client, user_id="alice", top_k=5)])

Memory db

python
from suprflo.integrations.agno import SuprfloMemoryDb

db = SuprfloMemoryDb(client, user_id="alice")
# pass `db` wherever Agno accepts a MemoryDb

Agno runs its own memory extraction and hands finished rows to the db, so upsert_memory writes with infer=False — the text is stored verbatim rather than re-extracted server-side.

Degraded methods

MemoryDb is modelled on a relational table; three methods have no counterpart in a hosted API:

MethodBehaviour
create()No-op — storage is server-managed.
table_exists()Always True.
drop_table() / clear()Delete every memory in scope. With no user_id set, that is the whole workspace.

If agno isn't installed, importing this module raises a clear ImportError telling you to pip install "suprflo[agno]".

The memory layer for AI agents.