Skip to content

Integrations ​

Connect your current app first: ChatGPT, Claude, or another host. Use the same Mnemoverse account across tools to share memories.

This catalog lists the available packages and endpoints. Their tool sets depend on the integration: see MCP tool availability and the separate REST/SDK references. A package version or public listing does not by itself verify installation in every host or plan.

The npm package, desktop bundle and Python SDK versions below come from the shared documentation facts registry. For other packages, follow the linked listing or release history for its current version; an installed copy may be older. Plugin source branches and packaged releases can differ, as explained in the host guides.

Connect directly ​

FormWhereAuthNotes
Remote MCP serverhttps://mcp.mnemoverse.com/mcpbrowser sign-in (OAuth 2.1 with PKCE), no keyStreamable HTTP. Any MCP client that can complete a browser sign-in. Guide
Local MCP server (npm)@mnemoverse/mcp-memory-server 0.17.1, MITmk_live_ API keynpx -y @mnemoverse/mcp-memory-server@latest. Source: github.com/mnemoverse/mcp-memory-server. Guide
REST APIhttps://core.mnemoverse.com/api/v1API keyWrite, read, feedback, stats from any language. Reference
Vercel AI SDK@mnemoverse/ai-sdk on npm, MIT; or tool() definitions over the REST APIAPI keycreateMnemoverseTools() on ai 5, 6 and 7, the MCP tools run in process; run live on 2026-10-03. Source: github.com/mnemoverse/ai-sdk. Guide
Python SDKmnemoverse 0.5.0 on PyPIAPI keypip install mnemoverse. Source: github.com/mnemoverse/mnemoverse-sdk-python. Guide

Plugins and extensions ​

HostPackageVersion sourceAuthInstall
ChatGPTMnemoverse MemoryPublished; current version on the public cardbrowser, no keyInstall plugin on the public card, then sign in. Setup and verification
Claude Codemnemoverse/claude-plugin, MITSource and release behaviorbrowser, no key/plugin marketplace add mnemoverse/claude-plugin, then /plugin install mnemoverse@mnemoverse. Details
Claude Desktopmnemoverse-memory.mcpb in each mcp-memory-server release0.17.1API key in the bundle settingsSettings, Extensions, Advanced settings, Install Extension. Details
Cursormnemoverse/cursor-plugin, MITRepository releasesbrowser, no keyClone into ~/.cursor/plugins/local/. Without a plugin: the keyless Add to Cursor link, or the extension below from Open VSX. Details
VS Code and Copilot ChatMnemoverse Agent Memory on the Marketplace, also on Open VSXCurrent version on Marketplace / Open VSXbrowser, no keycode --install-extension Mnemoverse.mnemoverse-vscode. Source: github.com/mnemoverse/mnemoverse-vscode. Details
Gemini CLImnemoverse/gemini-extension, MITRepository releasesbrowser, no keygemini extensions install https://github.com/mnemoverse/gemini-extension. Details
Kimi Codekimi.plugin.json in mnemoverse/claude-pluginLatest GitHub release by default; detailsbrowser, no key/plugins install https://github.com/mnemoverse/claude-plugin. Details

Agent Plugins 1.0.0 manifest ​

mnemoverse/claude-plugin carries plugin.json and mcp.json at its root, following the vendor-neutral Agent Plugins standard. Clients that read the standard install the same repository with their own command; the exact steps for each are on Install by host:

  • VS Code (Chat: Install Plugin From Source)
  • GitHub Copilot CLI
  • Kiro (Import power from GitHub)
  • Codex CLI (codex plugin marketplace add mnemoverse/claude-plugin); the Codex IDE extension uses a remote MCP entry; ChatGPT users install the published plugin
  • Grok Build (grok plugin install mnemoverse/claude-plugin --trust)
  • Cursor (local plugins folder)
  • Hermes Agent, OpenClaw, NanoClaw and OpenHands read the manifest and load the skills, but run no sign-in for a server that comes from a plugin; each section says how the server gets connected there

Standing rules for agents ​

agent-memory-discipline is a CC0 skill that tells an agent when to recall before acting and what is worth saving. It names no vendor and works with any memory backend. It is bundled as a skill in the Claude Code and Cursor packages above, as GEMINI.md in the Gemini CLI extension, and can be installed on its own; the technique behind it is explained in Make Your AI Agent Actually Use Persistent Memory.

Per-client guides ​