Vercel AI SDK Integration

@tekmemo/ai-sdk provides official helpers for integrating TekMemo memory into applications built with the Vercel AI SDK.

It allows your agents to automatically search memory for context and record new decisions or facts during a conversation.

Install

npm install @tekmemo/ai-sdk ai tekmemo @tekmemo/fs

Core Helpers

Helper	Purpose
buildRuntimeMemoryContext()	Compiles memory-aware context (core memory, notes, recall) for system prompts.
buildRuntimeMemoryToolDefinition()	Returns an AI SDK tool definition for runtime-based memory operations.
runRuntimeMemoryTool()	Executes runtime memory tool commands with scope enforcement.
createLocalAiSdkRuntime()	Creates a local runtime backed by a MemoryStore.

Example Usage

This example shows how to ground a generateText call in your project's local memory.

import { generateText } from "ai";
import { openai } from "@ai-sdk/openai";
import {
	buildRuntimeMemoryContext,
	buildRuntimeMemoryToolDefinition,
	createLocalAiSdkRuntime,
} from "@tekmemo/ai-sdk";
import { createNodeFsMemoryStore } from "@tekmemo/fs";

const store = createNodeFsMemoryStore({ rootDir: "./.tekmemo" });
const runtime = createLocalAiSdkRuntime({ workspace: store });

async function chat(userPrompt: string) {
	// 1. Build a memory-aware context
	// This reads core memory, notes, and performs recall search, then
	// compiles the results into a context block for the system prompt.
	const { text: system } = await buildRuntimeMemoryContext({
		runtime,
		access: { projectId: "my-project" },
		query: userPrompt,
		baseInstructions: "You are a helpful coding assistant.",
	});

	// 2. Generate text with memory tools
	// The model can now call tools to read core memory or save new notes.
	const result = await generateText({
		model: openai("gpt-4o"),
		system,
		prompt: userPrompt,
		tools: {
			memory: buildRuntimeMemoryToolDefinition({
				runtime,
				access: { projectId: "my-project" },
				allowWrites: true,
			}),
		},
	});

	return result.text;
}

Architectural Note

The AI SDK helpers are designed to be "pluggable." They rely on the MemoryStore and RecallStore interfaces, meaning they work equally well with local filesystem memory (@tekmemo/fs) or hosted cloud memory (@tekmemo/cloud-client).