XMemo: Memory OS for AI Agents

XMemo is a user-owned Memory OS for AI agents: persistent memory, project context and session continuity with agent identity, provenance and governed access, shared across AI clients through MCP, the CLI and a public REST API surface.

English · 简体中文 | Documentation · Hosted MCP

Quick start · What this repo contains · Beyond MCP · Integrations · Connection modes · Plugins · Commands · Versioning · Security

---

What this repository contains

• @xmemo/client: the official xmemo CLI for setup, diagnostics, behavior profiles, and memory commands.
• MCP distribution: hosted Streamable HTTP configuration (https://xmemo.dev/mcp) and the local xmemo-mcp stdio server.
• XMemo Skill & native integrations: skills/xmemo for agent platforms and the native plugin index (src/plugins/index.json).
• Marketplace and registry metadata: canonical descriptors for MCP Registry (server.json), LobeHub (lhm.plugin.json), and AI agents (context7.json).

The hosted service implementation is private.

Beyond MCP

MCP is one access surface of XMemo, not the product boundary. XMemo provides a persistent, governed memory operating layer across AI tools and agents:

• Persistent Memory: store, search, list, and soft-delete durable facts, preferences, and operational knowledge across sessions.
• Context Recall: retrieve focused, relevant context on demand with token-budget controls and document expansions.
• Session Continuity & Working State: restart snapshots and state save/restore so agents resume context across process restarts.
• Project Context & Decisions: track project-scoped facts, TODOs, and durable decisions across agent workflows.
• Agent Identity & Provenance: track authorship and origin per memory record via agent instance attribution without conflating credentials.
• Governed Access: scoped tokens, OAuth where the client supports it, soft/hard deletion and sensitive-memory handling (governance details).
• Native Integrations & Skill: dedicated integration for OpenClaw, Hermes, XMemo Skill (ClawHub), and DeepSeek DSH (available, released and pilot-tested; integration maturity Preview).
• Access Surfaces: unified via Model Context Protocol, the local CLI (xmemo), and the public REST API.

Preview: Cloud Skills; Dream (off by default); Teams (Business plan not yet available). Knowledge Bases are available where enabled for the account.

XMemo CLI

@xmemo/client is the official control plane for connecting AI tools to XMemo. It makes setup repeatable, keeps credentials out of project files, and gives every supported client a consistent path to durable, user-owned memory.

The package is deliberately small: the CLI runtime, safe client configuration, behavior profiles, XMemo skills, and marketplace metadata. Server code, databases, deployment files, logs, and internal operations remain outside the npm distribution.

Architecture

Package	@xmemo/client
Primary command	xmemo (alias: client)
Local MCP command	xmemo-mcp
Hosted MCP	https://xmemo.dev/mcp
Runtime	Node.js 20 or later
License	MIT

Why XMemo CLI

• One control plane — login, diagnostics, configuration, profiles, updates, and smoke checks share one predictable interface.
• Private by design — generated project configuration references a credential; it never embeds the credential value.
• Native where it matters — OpenClaw and Hermes use dedicated memory integrations instead of duplicating the same capability through MCP.
• Portable everywhere else — hosted Streamable HTTP MCP and local stdio cover modern editors, terminals, and agent runtimes.
• Safe automation — supported setup and removal paths offer preview, dry-run, or explicit confirmation before making changes.
• Small supply-chain surface — the npm package is governed by an explicit file allowlist and release provenance.

Quick start

Guided onboarding (xmemo init)

For an interactive first-run experience across account authentication, detected clients, agent behavior instructions, MCP server setup, skills, and plugins, run:

npm install -g @xmemo/client
xmemo init

Flags and options:

• xmemo init --dry-run: View the full onboarding plan without performing network calls or disk writes.
• xmemo init --yes: Automatically accept and apply all onboarding steps without interactive prompts.
• xmemo init --json: Emit structured JSON for the plan or result envelope.
• xmemo init --client <id>...: Restrict onboarding to specific clients (e.g., cursor, codex, claude-code).
• xmemo start: Alias for xmemo init with quick-start walkthrough steps.

Global installation exposes xmemo as the primary command, and also provides client and memory-os as aliases.

Manual step-by-step setup

xmemo account login
xmemo doctor
xmemo setup codex
xmemo status

Replace codex with your client. Preview a configuration before writing it:

xmemo setup cursor --dry-run

Running with npx

You can also run any CLI command directly without a global install via npx @xmemo/client <command>:

# Check version or health
npx @xmemo/client --version
npx @xmemo/client doctor

# Guided onboarding without global install
npx @xmemo/client init

# Install skill or run MCP stdio server
npx @xmemo/client skill install
npx @xmemo/client mcp serve

[!TIP] Start with xmemo init (or xmemo account login, xmemo doctor, and xmemo setup <client>). Hand-edit MCP configuration only when a client has no verified setup path.

How the commands fit together

The XMemo CLI architecture is built on four core design principles:

1. Unified Resource Grammar (xmemo <resource> <action>)

Every integration component is a first-class resource with predictable lifecycle actions:

Resource	Scope	Actions	Examples
mcp	MCP server connection configuration	install, remove, status	xmemo mcp install codex, xmemo mcp status
plugin	Host-native extension packages	install, remove, status, list, info	xmemo plugin install gemini-cli, xmemo plugin list
skill	Agent skill scripts & documentation	install, remove, status, update	xmemo skill install --client openclaw
profile	Markdown behavior steering instructions	install, remove, status, show	xmemo profile install cursor

• Composite commands: xmemo setup [<client>...], xmemo uninstall [<client>...], and xmemo status [<client>] orchestrate these resources in a single step according to the client's declarative profile.
• Backward-compatible aliases: Familiar commands such as xmemo mcp add (alias for mcp install), xmemo profile uninstall (alias for profile remove), and xmemo skill uninstall (alias for skill remove) remain fully functional and print a helpful one-line hint in interactive terminals.

2. Unified Target Resolver

When no client is explicitly passed, the CLI uses a deterministic three-tier precedence resolution model:

1. Explicit flag or argument: --client <id>, positional client argument, or --all.
2. Calling agent environment: Automatically identifies the calling agent runtime when running inside an agent session (e.g., CLAUDECODE / CLAUDE_CODE_ENTRYPOINT maps to claude-code, and CODEX_THREAD_ID / CODEX_SESSION_ID maps to codex).
3. Detected installed clients: Inspects local configuration paths and markers. If exactly one matching client is found, it is automatically selected; if multiple clients are found in interactive mode, an interactive picker is presented. The CLI never silently writes configuration to arbitrary unverified paths.

3. Plan, Confirm Once, Apply (PlanRunner)

Mutating commands follow a strict, atomic execution pattern:

1. Build Plan: Assemble an ordered sequence of actions across resources (e.g. plugin install followed by skill configuration).
2. Preview: Print the entire plan (including modified paths, commands, and unified diffs) to the terminal.
3. Confirm Once: Prompt [y/N] exactly once for the entire sequence. Re-running with --yes or -y bypasses the prompt; --dry-run displays the preview without mutation.
4. Apply Sequentially: Steps run in dependency order, stopping immediately upon first failure. Re-running an already configured client detects that all components are up to date and reports Nothing to do.

4. Declarative Client Registry

All client configurations, recipes, and capabilities are declared centrally in src/clients/registry.js. Command implementations are purely generic orchestrators with zero hardcoded client ID strings. Platforms can also be added dynamically at runtime via registerClient().

Supported integrations

Client	Recommended command	Connection
Codex	xmemo setup codex	Hosted MCP + behavior profile
Cursor	xmemo setup cursor	Hosted MCP + Bearer Token + behavior profile
Copilot CLI	xmemo setup copilot	Local authenticated proxy
Gemini CLI	xmemo setup gemini	Hosted MCP + OAuth
Antigravity	xmemo setup antigravity	Hosted MCP + OAuth
OpenClaw	xmemo setup openclaw	Native memory plugin + Skill
Hermes	xmemo setup hermes	Native memory provider
Kiro	xmemo setup kiro	Native HTTP OAuth; --auth key for API Key
Grok	xmemo setup grok	Hosted MCP
Other MCP clients	xmemo mcp config --client generic	Generated template

The client registry also covers Devin Desktop (formerly Windsurf), Cline, Continue, Claude Desktop, Claude Code, Kimi Code, Zed, JetBrains, OpenCode, Qwen, Trae, and compatible MCP hosts. Run xmemo mcp list for the current machine-readable catalog.

For VS Code users looking for the dedicated editor extension, see the yonro/xmemo-vscode repository. For Cursor users looking for the dedicated plugin, see the yonro/xmemo-cursor-plugin repository. For Claude users looking for the dedicated plugin, see the yonro/xmemo-claude-plugin repository.

Connection modes

Hosted MCP

The recommended universal path is the XMemo Streamable HTTP endpoint:

https://xmemo.dev/mcp

OAuth-capable clients complete authentication in the browser. Other clients reference XMEMO_KEY without copying its value into repository files.

Generic configuration shape:

{
  "mcpServers": {
    "XMemo": {
      "type": "streamable-http",
      "url": "https://xmemo.dev/mcp",
      "headers": {
        "Authorization": "Bearer ${XMEMO_KEY}"
      }
    }
  }
}

Client configuration keys differ; prefer xmemo setup <client> over copying this generic example directly.

Local stdio MCP

xmemo-mcp is the dedicated stdio entry point for marketplaces and clients that launch a local process. Safe discovery exposes 20 tools, three prompts, and two documentation resources without a token. Tool execution still requires authentication.

After a global installation:

xmemo-mcp

Install-free MCP configuration:

{
  "mcpServers": {
    "XMemo": {
      "command": "npx",
      "args": [
        "-y",
        "--package",
        "@xmemo/client@latest",
        "xmemo-mcp"
      ]
    }
  }
}

xmemo mcp serve is equivalent when the CLI is already installed.

Native integrations

OpenClaw and Hermes have dedicated memory providers. Their default setup avoids installing a second, duplicate XMemo tool surface:

• OpenClaw: installs the pinned plugin clawhub:@xmemo/openclaw-memory@1.0.18 without --force by default. Re-running setup gracefully detects existing installations; use --force to reinstall or overwrite.
• Hermes: installs the pinned provider package hermes-xmemo==1.1.3 via pip install without -U.
• Every install command prints the exact command before executing. Use --dry-run to preview actions without installing.

# Native OpenClaw plugin (clawhub:@xmemo/openclaw-memory@1.0.18) + XMemo Skill
xmemo setup openclaw

# Native Hermes memory provider (hermes-xmemo==1.1.3)
xmemo setup hermes

Add hosted MCP only when an explicit fallback is desired:

xmemo setup openclaw --with-mcp
xmemo setup hermes --with-mcp

Use --mcp-only to skip the native integration and install only the hosted MCP fallback.

XMemo Skill install

The CLI installs the verified XMemo Skill locally into agent skill folders or a target directory:

• Installs into client skill directories:
  • Claude Code: ~/.claude/skills/xmemo-memory (global) or .claude/skills/xmemo-memory (project with --project)
  • Codex: ~/.codex/skills/xmemo-memory
  • OpenClaw: ~/.openclaw/skills/xmemo-memory
  • All other 21 clients remain null until officially documented.
• Defaults to the pinned @xmemo/skill@1.1.35 release from npm and verifies tarball integrity (sha512 SRI) before extraction.
• Override version with --version <semver> or explicitly opt into the latest release via --version latest.
• For air-gapped or offline installations, install from a local directory or packed tarball with --from <dir|tgz> (optional --integrity <sha512>).
• Manage client skills with skill status, skill update, and skill remove.
• Preview actions without writing files using --dry-run.
• Safety & Consent:
  • Interactive install prompts [y/N] before writing (Enter, EOF, or empty input cancels) unless --yes is specified.
  • Existing installs refuse overwrite without --force; when --force is used, a backup is created in ~/.xmemo/backups/skills/<client>/ (outside the agent skills directory).
  • skill remove only removes verified XMemo skill directories, refusing foreign folders, and reports the preserved backup location.

# Install to agent skill folder (Claude Code global, Codex, or OpenClaw)
xmemo skill install --client claude-code
xmemo skill install --client codex
xmemo skill install --client openclaw

# Install OpenClaw skill globally (shared ~/.openclaw/skills)
xmemo skill install --client openclaw --global

# Install to project-level skill folder (Claude Code project: .claude/skills/xmemo-memory)
xmemo skill install --client claude-code --project

# Install for all detected supported clients
xmemo skill install --all

# Automatically resolve detected client(s) (or specify --client <id>)
xmemo skill install

# Install into a custom directory path
xmemo skill install --dir ./custom-skill-dir

# Non-interactive install (skips [y/N] prompt)
xmemo skill install --client codex --yes

# Replace existing installation (creates backup in ~/.xmemo/backups/skills/<client>/)
xmemo skill install --client codex --force --yes

# Update alias (equivalent to skill install --force)
xmemo skill update --client codex --yes

# Inspect installation status across clients
xmemo skill status
xmemo skill status --client codex
xmemo skill status --all --json

# Remove installed skill from an agent folder (refuses non-XMemo folders)
xmemo skill remove --client codex --yes
xmemo skill remove --client claude-code --project --yes

# Dry run preview
xmemo skill install --client codex --dry-run

Standalone curl & PowerShell installers

For environments without Node.js or @xmemo/client, XMemo provides standalone HTTPS installers at https://xmemo.dev/skill/install (POSIX sh) and https://xmemo.dev/skill/install.ps1 (PowerShell).

The installer is agent-aware and automatically resolves the correct target directory for your active agent:

# Claude Code: installs to ~/.claude/skills/xmemo-memory
curl -fsSL https://xmemo.dev/skill/install | XMEMO_SKILL_AGENT=claude-code sh

# Codex: installs to ${CODEX_HOME:-$HOME/.codex}/skills/xmemo-memory
curl -fsSL https://xmemo.dev/skill/install | XMEMO_SKILL_AGENT=codex sh

# OpenClaw: install via OpenClaw CLI
openclaw skills install @xmemo/xmemo --version 1.1.35

# Windows (PowerShell):
# $env:XMEMO_SKILL_AGENT="claude-code"; irm https://xmemo.dev/skill/install.ps1 | iex
# $env:XMEMO_SKILL_AGENT="codex"; irm https://xmemo.dev/skill/install.ps1 | iex

[!TIP] Prompt for your AI Agent (copy and paste directly into Claude Code, Codex, or another agent):

Install the standalone XMemo Skill into your skills directory: run the installer from https://xmemo.dev/skill/install (on Windows, https://xmemo.dev/skill/install.ps1) with XMEMO_SKILL_DIR set to your skills directory plus /xmemo-memory. Run the doctor command it prints to verify the installation, then reload your skills and follow the Skill's first-run steps. Follow your normal safety checks and do not use elevated privileges.

Target resolution precedence (first match wins):

1. XMEMO_SKILL_DIR: Installs to the specified directory.
2. XMEMO_SKILL_AGENT=claude-code|codex: Installs to the explicit agent's skills directory (openclaw redirects to openclaw skills install xmemo).
3. Auto-detection: Automatically detects Claude Code (CLAUDECODE=1) or Codex (CODEX_THREAD_ID, CODEX_SESSION_ID, or CODEX_HOME).
4. Home directory discovery: If only ~/.claude or only ~/.codex exists in HOME, selects that agent.
5. Fallback: Installs to ./xmemo-skill with a warning on stderr explaining that AI agents will not automatically load the skill from this directory.

Safety & Replacement:

• Refuses to overwrite existing installations unless XMEMO_SKILL_FORCE=1 is provided.
• When replacing, moves the previous installation to ~/.xmemo/backups/skills/<agent>/<name>-<timestamp> (safely outside agent skill search paths).
• After installation, prints the absolute install path, the verification doctor command (node <path>/scripts/xmemo-skill.mjs doctor --anonymous), and a prompt to reload your agent.

Agent plugins

The CLI provides a curated, static index of verified agent plugins shipped directly in @xmemo/client. Each entry contains a pinned version, release tag, and exact Git commit SHA resolved at release time.

ℹ️ Strict Separation Rule: xmemo plugin installs agent plugins only (e.g. @xmemo/openclaw-memory). Skills are installed exclusively via xmemo skill install (e.g. @xmemo/xmemo).

Plugin ID	Platform / Agent	Kind	Status	Integration
openclaw	OpenClaw	native-cli	Stable	openclaw plugins install clawhub:@xmemo/openclaw-memory@1.0.18 (auto-prompts update if already installed)
hermes	Hermes Agent	native-cli	Stable	hermes plugins install xmemo (fallback: python -m pip install hermes-xmemo==1.1.3)
claude-code	Claude Code	git-dir	Preview	Pinned Git clone verified against commit 5d0d280 (defaults to ~/.xmemo/plugins/claude-code)
cursor	Cursor	marketplace	Preview	Cursor Marketplace plugin
gemini-cli	Gemini CLI	native-cli	Preview	gemini extensions install https://github.com/yonro/xmemo-gemini-cli --ref 39e25b185b5157490d1683e4ca8c5c5fb1312a88
kiro	Kiro	manual	Preview	Steering rules & Power integration
vscode	VS Code	manual	Preview	VS Code extension manual steps (pending marketplace publication)
deepseek-dsh	DeepSeek DSH	native-cli	Preview	dsh plugin --profile <name> add dsh-xmemo (requires --profile)
chatgpt-codex	ChatGPT / Codex	marketplace	Preview	ChatGPT & Codex extension
cindy	Cindy	manual	Preview	Native agent memory integration
codex	Codex	mcp	Preview	Dedicated MCP configuration (xmemo setup codex)

Commands:

# List available plugins (excluding legacy entries)
xmemo plugin list

# Include legacy plugins
xmemo plugin list --all

# View plugin details and verification metadata
xmemo plugin info <id>

# Preview install plan without executing
xmemo plugin install <id> --dry-run

# Install with explicit confirmation (prompts [y/N] by default)
xmemo plugin install <id>

# Non-interactive install
xmemo plugin install <id> --yes

# Specify profile for deepseek-dsh
xmemo plugin install deepseek-dsh --profile default --yes

# Specify custom target directory for git-dir plugins
xmemo plugin install claude-code --yes --dir ~/.custom-plugins/claude-code

# Open plugin documentation or marketplace in browser
xmemo plugin install <id> --open

# Check installation status
xmemo plugin status [<id>]