kronvex 0.2.0

Persistent memory for AI agents — Kronvex Python SDK

Kronvex Python SDK

Persistent memory for AI agents. Three endpoints, one API key, production-ready.

Installation

pip install kronvex

Quick start

from kronvex import Kronvex

kx = Kronvex("kx_your_api_key")
agent = kx.agent("your-agent-id")

# Store a memory
agent.remember("User prefers concise answers", memory_type="preference")

# Recall relevant memories
memories = agent.recall("what does the user prefer?", top_k=5)
for m in memories:
    print(f"[{m['score']:.2f}] {m['content']}")

# Inject context into your prompt
context = agent.inject_context("How should I respond?")
# → "Relevant memories:\n- User prefers concise answers\n..."

Async support

import asyncio
from kronvex import AsyncKronvex

async def main():
    async with AsyncKronvex("kx_your_api_key") as kx:
        agent = kx.agent("your-agent-id")
        await agent.remember("User is based in Paris", memory_type="semantic")
        memories = await agent.recall("where is the user?")

asyncio.run(main())

With LangChain

from kronvex import Kronvex
from langchain_openai import ChatOpenAI

kx = Kronvex("kx_your_api_key")
agent = kx.agent("your-agent-id")

def chat(user_message: str, session_id: str) -> str:
    # Inject past context
    context = agent.inject_context(user_message, session_id=session_id)

    llm = ChatOpenAI(model="gpt-4o")
    response = llm.invoke([
        {"role": "system", "content": f"You are a helpful assistant.\n\n{context}"},
        {"role": "user", "content": user_message},
    ])

    # Store the exchange
    agent.remember(f"User: {user_message}", memory_type="episodic", session_id=session_id)
    agent.remember(f"Assistant: {response.content}", memory_type="episodic", session_id=session_id)

    return response.content

API reference

Kronvex(api_key, *, base_url, timeout)

Method	Description
.agent(agent_id)	Get an Agent handle
.create_agent(name)	Create a new agent
.list_agents()	List all agents

Agent

Method	Description
.remember(content, *, memory_type, session_id, ttl_days, pinned, metadata)	Store a memory
.recall(query, *, top_k, memory_type, session_id, threshold)	Semantic search
.inject_context(message, *, top_k, session_id)	Get prompt-ready context block
.sessions()	List session IDs
.memories(*, session_id, memory_type, limit, offset)	List stored memories
.delete_memory(memory_id)	Delete one memory
.clear()	Delete all memories

Links

• Dashboard
• API Docs
• Node.js SDK