mubit-agno 0.5.2

Agno memory and toolkit integration backed by MuBit memory engine

mubit-agno

Agno memory and toolkit integration backed by the MuBit memory engine.

Installation

pip install mubit-agno[agno]

Quick Start

Memory DB — Persistent Agent Memory

Use MubitMemoryDb as the database backend for Agno's built-in memory system:

from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.memory.v2.memory import Memory
from mubit_agno import MubitMemoryDb

agent = Agent(
    model=OpenAIChat(id="gpt-4o"),
    memory=Memory(db=MubitMemoryDb(
        api_key="mbt_...",
        session_id="user-session-1",
        user_id="user-42",
    )),
    enable_agentic_memory=True,
)

agent.run("Remember that I prefer concise answers")
agent.run("What are my preferences?")  # Recalls from MuBit

Toolkit — Direct Memory Tools

Give agents LLM-callable tools for fine-grained memory control:

from agno.agent import Agent
from agno.models.openai import OpenAIChat
from mubit_agno import MubitToolkit

agent = Agent(
    model=OpenAIChat(id="gpt-4o"),
    tools=[MubitToolkit(
        api_key="mbt_...",
        session_id="research-run-1",
    )],
)

# Agent can now call mubit_remember, mubit_recall, mubit_reflect,
# mubit_get_context, mubit_checkpoint, mubit_diagnose, mubit_memory_health
agent.run("Store a lesson: always validate input before processing")
agent.run("What lessons have we learned?")

Convenience Wrapper — Full Integration

MubitAgnoMemory bundles both surfaces and adds MAS extensions:

from agno.agent import Agent
from agno.models.openai import OpenAIChat
from agno.memory.v2.memory import Memory
from mubit_agno import MubitAgnoMemory

mubit = MubitAgnoMemory(
    api_key="mbt_...",
    session_id="crew-run-1",
    user_id="user-42",
)

agent = Agent(
    model=OpenAIChat(id="gpt-4o"),
    memory=Memory(db=mubit.as_memory_db()),
    tools=[mubit.as_toolkit()],
)

# MAS coordination
mubit.register_agent("researcher", role="researcher",
                      read_scopes=["fact", "lesson"],
                      write_scopes=["trace", "lesson"])
mubit.checkpoint("Phase 1", "Research complete")
mubit.record_outcome("task-1", "success", rationale="All sources verified")
strategies = mubit.surface_strategies()

Extended Features

The MubitAgnoMemory wrapper provides full MAS capabilities:

• Context: get_context(query) — pre-assembled memory context
• Reflection: reflect() — extract lessons from evidence
• Lessons: lessons() — list and filter learned lessons
• Checkpoints: checkpoint(label, snapshot) — durable snapshots
• Outcomes: record_outcome(ref, outcome) — RL-style feedback
• Step Outcomes: record_step_outcome(step_id) — per-step rewards
• Strategies: surface_strategies() — pattern discovery
• Agent Registration: register_agent(id, role, scopes) — MAS setup
• Handoffs: handoff(from, to, content) — agent coordination
• Feedback: feedback(handoff_id, verdict) — async evaluation
• Archive: archive(content, kind) — exact reference storage
• Dereference: dereference(ref_id) — fetch archived content
• Diagnostics: diagnose(error) — failure analysis
• Health: memory_health() — quality assessment

Environment Variables

Variable	Description	Default
MUBIT_ENDPOINT	MuBit server URL	http://127.0.0.1:3000
MUBIT_API_KEY	MuBit API key	(empty for local dev)

License

Apache-2.0