Stash

Stash gives AI tools persistent memory so preferences, plans and working context can carry into later conversations. It's for people switching between AI tools and developers building agents that need information to survive between sessions.

• Shared context: The hosted service connects Claude Desktop, Claude Code, Cursor, OpenCode and Replit through remote MCP, so compatible tools can recall information from the same memory store.
• Memory editing: In the hosted service, ask a connected AI to inspect, update or remove saved memories; memory is private by default.
• Self-hosted storage: The open-source project stores episodes, facts and working context in Postgres with pgvector and includes an MCP server for compatible agents.
• Knowledge consolidation: The self-hosted project processes observations into facts and relationships, with support for contradictions, goals and recurring failure patterns.

The hosted service requires Google sign-in and keeps memories in Stash's cloud. Saving and recalling information depends on the connected AI tool using Stash; it doesn't automatically retain every conversation.

The self-hosted project uses Apache 2.0 and runs as a Go binary, with Docker deployment also supported. You control its storage, but it needs embedding and reasoning models through an OpenAI-compatible API, including Atlas Cloud. If those models run through an external provider, their processing sends data to that provider.

The hosted service has a separate codebase, and its feature set differs from the self-hosted project.