The AI Memory SDK for Every Application

Squish gives your AI apps persistent memory — remember user context, past conversations, and learned preferences across sessions.

Drop-In Memory for Any AI App

Squish integrates as an MCP server or direct SDK. Your app gains remember, recall, forget, and inspect capabilities with three commands and zero code changes.

39 Ops/Second at the Edge

6.6ms embedding generation, 6.1ms semantic search, 39 ops/sec sustained throughput. All running locally with SQLite — no external APIs or cloud dependencies.

From Prototype to Production

Start free with the local runtime. Add cloud sync, team memory, and managed infrastructure when your application grows. The same SDK works at every scale.

Built for MCP and Beyond

Natively supports the Model Context Protocol for agent integration. Also works as a standalone SDK for custom apps, LangChain agents, and Python or Node.js workflows.

Frequently Asked Questions

What SDK languages do you support?

Squish works with any MCP-compatible client. Native support for Node.js, Python, and CLI workflows. LangChain integration available via the MCP adapter.

Can I use the SDK without MCP?

Yes. Use Squish directly as a local SDK with the squish CLI or programmatic API for custom application integration.

Is the SDK free for commercial applications?

Yes. The Squish runtime is MIT-licensed open source. Use it in commercial applications with no restrictions.

Does the SDK support cloud sync?

Yes. Squish Cloud syncs memory across devices and team members. Enable with one command after installing the SDK.

Start Building Free

Squish documentation