01
Shared memory layer
A common memory layer spans Copilot, Claude, Codex, Gemini, IDEs, CLIs, and conversations without confusing owners or sessions.
Open Beta · Agent Memory Control Plane
XMemo
One memory layer for every agent — every entry knows who wrote it
XMemo is Yonro's open beta Agent Memory Control Plane: a shared, identity-aware memory layer across Copilot, Claude, Codex, Gemini, MCP clients, CLIs, IDEs, and conversations. Registration is open at xmemo.dev, with discovery documents, MCP and REST endpoints, local client setup through @xmemo/client, and governed controls so agents can remember, recall, and attribute context safely.
Core Capabilities
Live agent memory control plane
XMemo is live at xmemo.dev with open registration, hosted MCP, REST, agent discovery, OAuth-ready ChatGPT integration, token-based CLI/MCP clients, auditability, and production rate limiting.
02
Identity-aware attribution
Every entry carries agent identity and local instance attribution, so teams know which agent wrote or used each memory.
03
Recall across tools
Agents can search, recall, remember, update, forget, redact, restore, and explain memory through MCP tools and REST endpoints.
04
Hosted beta on Azure
The public beta runs on Azure infrastructure and is supported by Microsoft for Startups, with production service-edge controls.
05
MCP and client onboarding
Discovery endpoints provide client configuration for Codex, ChatGPT, Gemini CLI, Copilot CLI, Claude Code, Cursor, Cline, Continue, Windsurf, and generic MCP clients.
06
REST, MCP, and CLI
Use the hosted MCP endpoint, REST API, or the npm client: npm install -g @xmemo/client.
07
Secure control-plane boundary
Discovery is read-only, tokens stay out of URLs and source control, and agents cannot run migrations, rotate tokens, or alter server configuration.
08
Governed memory lifecycle
Recall feedback, eval judgments, lifecycle runs, version history, conflict resolution, forget/redact, reminders, decisions, timelines, and audit queries keep memory reviewable.
Use Cases
Where XMemo is used
Agent external brain
Existing agents do not need to implement memory storage and recall themselves; they connect XMemo through MCP, REST, CLI, or official client profiles.
Session recovery
Project facts, user corrections, decisions, and deployment state can be recalled at the next task start instead of being rediscovered.
Shared team memory
Teams can store scoped knowledge that multiple agents reuse without mixing unrelated owners, workspaces, projects, or local agent instances.
MCP and CLI integrations
Codex, ChatGPT, Gemini CLI, Copilot CLI, Claude Code, Cursor, Claude Desktop, Cline, Continue, Windsurf, and generic MCP clients can be configured from XMemo discovery endpoints.
Knowledge governance
Versioning, conflict detection, forget/redact, audit logs, recall feedback, and token boundaries make agent memory reviewable instead of opaque.
Beta product adoption
The open beta is available at xmemo.dev with registration, login, hosted MCP, product docs, trust resources, privacy, terms, and support.
Integration path
How teams connect XMemo
- 01Open xmemo.dev and create a product account; registration is available now
- 02Install the CLI with npm install -g @xmemo/client or configure an MCP client from the discovery endpoints
- 03Set XMEMO_KEY and optional XMEMO_AGENT_ID / XMEMO_AGENT_INSTANCE_ID for local agent attribution
- 04Connect Codex, ChatGPT, Gemini CLI, Copilot CLI, Claude Code, Cursor, Claude Desktop, Cline, Continue, Windsurf, or a generic MCP client
- 05Use MCP or REST tools to remember, recall, update, forget, redact, restore, and explain memory
- 06Review trust, privacy, terms, support, audit, and lifecycle controls before production workflows
Connect your agents to XMemo
Visit the XMemo open beta or bring a concrete agent, workflow, or integration target. Yonro can help connect MCP clients, define attribution, and shape memory boundaries for practical production use.