XMemo - Persistent user-owned memory for AI agents

[yunze7373]

Display Name: XMemo

Category: Agent Skills

Primary Link: https://github.com/yonro/memory-os-cli

Author Name and Link: yonro

License: UNLICENSED (proprietary hosted service, CLI is open)

Description: XMemo gives Claude Code persistent, user-owned memory across sessions. Save decisions, recall project context, manage TODOs, and govern memory lifecycle through a hosted MCP endpoint at https://xmemo.dev/mcp. Supports OAuth and bearer token auth.

Validate Claims: Install via "npx skills add yonro/memory-os-cli" then ask Claude to "search XMemo for project conventions". The skill connects to the hosted MCP endpoint and returns memory results or a clear auth prompt.

Specific Task: Ask Claude: "Remember in XMemo: this project uses TypeScript strict mode" — it should save a memory via the XMemo MCP server and confirm the save.

[Keesan12]

Persistent memory is useful, but the part that usually decides whether people trust it is memory governance, not retrieval quality.

The checks I would want from a tool like this are:

• explicit scope (project, workspace, personal) on every write
• confidence / source tagging so recalled memories do not all look equally authoritative
• expiry or review semantics for operational facts that can drift
• a clean way to inspect "why did the agent believe this" when a memory starts steering behavior

That is where memory tools stop feeling like a neat demo and start feeling safe inside long-running agent work. Without that, persistence can turn one mistaken assumption into a very durable one.