Kronvex Node.js SDK

Persistent memory for AI agents. Works with any LLM framework — OpenAI, LangChain, Vercel AI SDK, etc.

Installation

npm install kronvex
# or
pnpm add kronvex
# or
yarn add kronvex

Quick start

import { Kronvex } from "kronvex";

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

// Store a memory
await agent.remember("User prefers concise answers", { memory_type: "preference" });

// Recall relevant memories
const memories = await agent.recall("what does the user prefer?", { top_k: 5 });
memories.forEach(m => console.log(`[${m.score?.toFixed(2)}] ${m.content}`));

// Inject context into your prompt
const context = await agent.injectContext("How should I respond?");

With Vercel AI SDK

import { Kronvex } from "kronvex";
import { openai } from "@ai-sdk/openai";
import { streamText } from "ai";

const kx = new Kronvex(process.env.KRONVEX_API_KEY!);

export async function POST(req: Request) {
  const { messages, agentId, sessionId } = await req.json();
  const agent = kx.agent(agentId);
  const lastMessage = messages.at(-1)?.content ?? "";

  // Get context from past sessions
  const context = await agent.injectContext(lastMessage, { session_id: sessionId });

  const result = streamText({
    model: openai("gpt-4o"),
    system: `You are a helpful assistant with memory.\n\n${context}`,
    messages,
    onFinish: async ({ text }) => {
      // Store the exchange
      await agent.remember(lastMessage, { memory_type: "episodic", session_id: sessionId });
      await agent.remember(text, { memory_type: "episodic", session_id: sessionId });
    },
  });

  return result.toDataStreamResponse();
}

With LangChain

import { Kronvex } from "kronvex";
import { ChatOpenAI } from "@langchain/openai";
import { HumanMessage, SystemMessage } from "@langchain/core/messages";

const kx = new Kronvex(process.env.KRONVEX_API_KEY!);
const agent = kx.agent(process.env.KRONVEX_AGENT_ID!);

async function chat(userMessage: string, sessionId: string): Promise<string> {
  const context = await agent.injectContext(userMessage, { session_id: sessionId });

  const llm = new ChatOpenAI({ model: "gpt-4o" });
  const response = await llm.invoke([
    new SystemMessage(`You are a helpful assistant.\n\n${context}`),
    new HumanMessage(userMessage),
  ]);

  // Store for next time
  await agent.remember(`User: ${userMessage}`, { session_id: sessionId });
  await agent.remember(`You: ${response.content}`, { session_id: sessionId });

  return String(response.content);
}

API reference

new Kronvex(apiKey, options?)

| Option | Default | Description | |--------|---------|-------------| | baseUrl | https://api.kronvex.io | API base URL | | timeout | 30000 | Request timeout in ms |

| Method | Returns | Description | |--------|---------|-------------| | .agent(agentId) | Agent | Get an Agent handle | | .createAgent(name, description?) | Promise<Agent> | Create a new agent | | .listAgents() | Promise<AgentData[]> | List all agents |

Agent

| Method | Returns | Description | |--------|---------|-------------| | .remember(content, options?) | Promise<Memory> | Store a memory | | .recall(query, options?) | Promise<Memory[]> | Semantic search | | .injectContext(message, options?) | Promise<string> | Get prompt context block | | .sessions() | Promise<Session[]> | List session IDs | | .memories(options?) | Promise<Memory[]> | List stored memories | | .deleteMemory(memoryId) | Promise<void> | Delete one memory | | .clear() | Promise<{deleted: number}> | Delete all memories |

Links

• Dashboard
• API Docs
• Python SDK