@tekmemo/server

@tekmemo/server is the Hono-based API layer for self-hosted TekMemo memory servers.

Root import

Use the root import for runtime-neutral app construction:

import { createTekMemoServer } from "@tekmemo/server";

The root import avoids Node-only dependencies so Cloudflare Workers can use the shared Hono app.

Node import

Use the Node subpath for Node/Docker/Railway/Fly deployments:

import {
	createNodeTekMemoObjectStore,
	createNodeTekMemoQueue,
	createNodeTekMemoStore,
} from "@tekmemo/server/node";

The Node runtime supports:

• Postgres memory persistence.
• pgvector extension-ready schema.
• Postgres-backed job queue.
• S3-compatible object storage.
• Local filesystem object storage for development.

Self-hosting with Node

The recommended Node deployment target is apps/self-host-node in the TekMemo repository.

It runs as one API process plus one worker process:

# API server
npm run server

# Background worker
npm run worker

This works with Railway, Fly.io, Render, Coolify, Northflank, and similar hosts.

Self-hosting with Docker Compose

apps/self-host-docker provides a Docker Compose setup that starts:

• tekmemo-api
• tekmemo-worker
• postgres with pgvector
• minio for S3-compatible object storage

cp .env.example .env
docker compose up --build

Connecting a client

Use @tekmemo/cloud-client against your self-hosted base URL:

import { createTekMemoCloudClient } from "@tekmemo/cloud-client";

const client = createTekMemoCloudClient({
	baseUrl: "http://localhost:8787/api/v1",
	apiKey: process.env.TEKMEMO_API_KEY,
	defaultProjectId: "default",
});