Skip to content

API documentation

New here? Start with the Quick start, then the API reference. Below, the pages in this directory, A–Z; the worked integrations are listed separately under Use cases.

  • API Reference: REST API reference for Mnemoverse memory: write, read, query, feedback, consolidate, stats, and delete endpoints, plus auth, error codes, and rate limits.
  • ChatGPT Integration: Give ChatGPT persistent memory with Mnemoverse: MCP connector with browser sign-in on personal accounts, or Custom GPT Actions with an API key in a Business/Enterprise workspace.
  • Claude Code: Claude Code (CLI) persistent memory via the Mnemoverse MCP server: one command, plus the tools Claude calls each session. See Claude Desktop, Web & Mobile too.
  • Claude Desktop, Web & Mobile: Add Mnemoverse memory to Claude Desktop, claude.ai, and mobile — one-click OAuth or a local config file. Manual setup today; directory listing rolling out.
  • Cursor: Add persistent memory to Cursor via MCP: one-click install or ~/.cursor/mcp.json with an API key, or keyless OAuth. Write once, recall in Claude Code or ChatGPT.
  • Cursor, VS Code & Windsurf: Cursor, VS Code, and Windsurf persistent memory via Mnemoverse — pick your editor: keyless extension, one-click Cursor install, or mcp.json with one key.
  • Getting Started: Free Mnemoverse API key, no credit card: store, query, and give feedback on your first memory in two minutes with curl against the REST API — no SDK needed.
  • MCP Server: The @mnemoverse/mcp-memory-server npm package gives any MCP client persistent memory via ten tools: five for your own memory, four for shared rooms, and one for stored secret aliases.
  • Make Your AI Agent Actually Use Persistent Memory: Get your AI agent to actually use Mnemoverse memory — the standing-instruction technique for CLAUDE.md, copilot-instructions, AGENTS.md, and editor rules.
  • Persistent Memory for AI Agents: Learns From Outcomes: Persistent memory API for AI agents. Hebbian associations + prediction-error feedback (Rescorla-Wagner-style), not vector search. Free tier, 1,000 queries/day.
  • Pricing: Mnemoverse pricing compared: a free tier, Pro at $29/mo and Team at $149/mo, plus Enterprise — with daily query, atom-storage, and rate-limit caps per plan.
  • Python SDK: Python SDK for the Mnemoverse memory API: pip install mnemoverse, then sync or async clients for write, batch write, read, feedback, stats, and health calls.
  • Quick Setup: Add Mnemoverse persistent memory to any AI tool: local stdio MCP (API key or keyless sign-in) or remote OAuth 2.1 + PKCE — configs, REST fallback, verify step.
  • Remote MCP Server (one-click sign-in): Connect Mnemoverse memory via the hosted remote MCP server — one-click OAuth, no API key. Streamable HTTP at mcp.mnemoverse.com/mcp, memory and room tools.
  • Rooms — Shared Memory Across Accounts and Agents: Shared memory for agent teams (beta): put xroom:<room_id> in the domain field of /memory/write and /memory/read — membership-checked, billed to caller.
  • Security: Mnemoverse security: per-organization tenant isolation, SHA-256 hashed API keys, constant-time comparison, HTTPS transport, no training on data, rate limits.
  • VS Code & Copilot Chat: Add persistent memory to VS Code and Copilot Chat: keyless Mnemoverse extension with browser sign-in, remote OAuth URL, or .vscode/mcp.json with an API key.
  • Windsurf (Devin Desktop): Add persistent memory to Windsurf (Devin Desktop) via MCP: mcp_config.json with an API key. Write once, recall in Claude Code, Cursor, or ChatGPT.