Geniffy for LlamaIndex

Give a LlamaIndex agent a memory of each of your users. Add one memory block, and before the agent answers, LlamaIndex puts what is known about the user into the system message, each line with where it came from. After the run, save the exchange. When nothing is known, the model is told so, and says so instead of guessing.

pip install llama-index-memory-geniffy

Set GENIFFY_API_KEY from API keys in the Geniffy app. Keep it on your server.

The memory block

from llama_index.core.agent.workflow import FunctionAgent
from llama_index.core.memory import Memory
from llama_index.llms.anthropic import Anthropic
from llama_index.memory.geniffy import GeniffyMemoryBlock

agent = FunctionAgent(llm=Anthropic(model="claude-opus-5-5"), system_prompt="You are a helpful assistant.")


async def chat(user_id: str, message: str) -> str:
    block = GeniffyMemoryBlock(space=f"user_{user_id}")       # the user from your own sign-in
    memory = Memory.from_defaults(session_id=f"user_{user_id}", memory_blocks=[block])
    reply = await agent.run(user_msg=message, memory=memory)
    await block.remember(message, reply)                      # the exchange, saved to that user's space
    return str(reply)

• Before the agent answers, what is known that bears on the user's latest message goes in the system message, inside LlamaIndex's <memory> section. In a run with tools, Geniffy is asked once, not once per step.
• remember() saves the exchange. LlamaIndex only hands a block the messages that overflow its short-term history, which most conversations never reach, so each exchange is saved this way instead. To save the overflow too, pass accept_short_term_memory=True and don't call remember(), or an exchange is saved twice.
• If Geniffy can't be reached, the agent goes on without memory, and on_error hears about it.

Options: instructions to change what the model is told about the memory, and client to share your own geniffy.AsyncGeniffy. Without one, a client is made for each event loop and shared by every block.

Tools

To let the agent decide when to look something up or save it:

from llama_index.memory.geniffy import geniffy_tools

agent = FunctionAgent(llm=llm, tools=geniffy_tools(space=f"user_{user_id}"))   # recall and remember

Make the tools per user (per request), so the model never sees or chooses whose memory it reads.

One space per user

space is required: the user this is for. Each space is a memory of its own, and nothing else can read it. space=None is your own memory, never your users' data, and a blank space is refused, so a user with no id can't end up in it. When a user deletes their account, forget them with AsyncGeniffy().forget_space(...) from the geniffy SDK.

Develop

pip install -e ".[test]" && pytest     # through LlamaIndex's own FunctionAgent and Memory, with a scripted LLM

Security

Report a vulnerability to ops@geniffy.com, not in a public issue. See the security policy.