Integrations
Every AI you use.
One memory.
Each AI keeps its own memory, if any. MemoryRouter keeps your conversational memory in one place, so you can bring your context into chat assistants, coding agents, and the apps you build. Choose your tool to get connected.
Save my memories01 / 03 entries
Chat assistants
Your context, wherever the conversation happens.
ChatGPT
Developer mode connectorSearch your shared memory and save the details you choose from ChatGPT.
https://mcp.memoryrouter.ai/mcpClaude and Cowork
Connector + optional skillsBring your memory into Claude conversations and Cowork tasks, with approved saves.
https://mcp.memoryrouter.ai/mcpClaude connectors
MCP in ClaudeConnect a memory vault directly through Claude’s custom connector settings.
https://mcp.memoryrouter.ai/mcp
02 / 13 entries
Coding agents
Keep the decisions. Pick up the project.
Claude Code
Native hooksRecall project context and capture conversations across sessions and compaction.
npx -y memoryrouter-claude initCodex
Native hooks + OAuthGive Codex persistent project memory with automatic recall and turn capture.
pbpaste | npx -y memoryrouter-codex init --scope user --auth oauth --api-key-stdinOpenClaw
Native pluginCarry conversational memory across agent sessions, channels, and model switches.
openclaw plugins install npm:mr-memoryDeepSeek Harness
Native pluginShare memory across dsh’s Web UI, messaging gateways, and headless sessions.
dsh plugin --profile <your-profile> add dsh-memoryrouterHermes Agent
Native pluginAutomatically capture clean conversations and recall relevant context across Hermes sessions.
pip install hermes-memoryrouter hermes-memoryrouter install hermes config set memory.provider memoryrouterNemoClaw
Sandbox setupConnect a sandboxed NemoClaw agent to your MemoryRouter vault.
npx @memoryrouter/nemoclaw setup --sandbox my-assistant --api-key <your-memory-key>Cursor
Remote MCP + OAuthAdd MemoryRouter as a remote MCP server. OAuth built in, no API key in any config file.
https://mcp.memoryrouter.ai/mcpWindsurf
Remote MCP + OAuthCascade connects to MemoryRouter over MCP with OAuth built in.
https://mcp.memoryrouter.ai/mcpCline
Remote MCP + OAuthRemote MCP server with OAuth 2.1 support in cline_mcp_settings.json.
https://mcp.memoryrouter.ai/mcpContinue
Remote MCP + OAuthAdd a remote MCP server to config.yaml and connect from agent mode.
https://mcp.memoryrouter.ai/mcpAider
OpenAI-compatible proxyNo MCP support and no plugin system. Point OPENAI_API_BASE at the MemoryRouter proxy instead.
https://api.memoryrouter.ai/v1Sourcegraph Cody
Local MCP bridge (Enterprise)Bridge to MemoryRouter’s remote MCP server with mcp-remote. Requires Sourcegraph Enterprise with agentic context fetching enabled.
https://mcp.memoryrouter.ai/mcpZed
Remote MCP + OAuthAdd MemoryRouter as a remote context server. Zed handles OAuth automatically.
https://mcp.memoryrouter.ai/mcp
03 / 08 entries
Frameworks and protocols
Build memory into the tools you make and the stack you already use.
Any MCP client
Remote MCP + OAuthConnect clients that support remote Streamable HTTP MCP and OAuth.
https://mcp.memoryrouter.ai/mcpLangChain
Python packageGive your agents retain and recall tools backed by user-scoped memory.
pip install langchain-memoryrouterLangGraph
Python packageGive every user of your agent a private, persistent memory vault with tools, graph nodes, or a BaseStore.
pip install langchain-memoryrouterProxy API
OpenAI-compatible APIAdd memory to model requests through a drop-in API base URL.
https://api.memoryrouter.ai/v1MemoryRouter CLI
Command lineUpload files, search memories, and manage your vault from the terminal.
npm install -g @memoryrouter/cliOpen WebUI
Provider + history uploadConnect your chat UI to memory and backfill history with the upload function.
https://api.memoryrouter.ai/v1Local inference
Prepare + ingest APIKeep model calls in your own stack while MemoryRouter handles retrieval and storage.
POST /v1/memory/prepareSupported models
Provider compatibilityExplore the model and provider options available through the memory proxy.
04 / 04 entries
Data and imports
Bring the history. Keep the continuity.
ChatGPT to Claude memory transfer
Guided transferReview what ChatGPT can currently recall, save it, and pick up in Claude.
ChatGPT history import
Local archive importerImport your official ChatGPT export with a preview, explicit approval, and a receipt.
memoryrouter chatgpt-import inspect /path/to/export.zipAgent history import
Connector CLIBring local Claude Code, Codex, and OpenClaw conversation history into your vault.
npx memoryrouter-claude import --instructions memoryrouter-codex import --instructions openclaw mr import --instructionsVault transfer and graft
Memory APICopy conversation memories or selected matches between vaults without changing the source.
POST /v1/memory/transfer
One account. Your memory.
Change tools.
Not your starting point.
Connect the same vault in each tool. Chat connectors search and save when asked; hooks and API integrations handle memory through their own workflows.
Which products give developers an API for shared, persistent conversation memory? Three kinds: memory APIs you call from your own code (MemoryRouter, Mem0, Zep, Supermemory), MCP memory servers that any compatible client can call, and OpenAI-compatible memory proxies that add recall to every model request. MemoryRouter covers all three from one vault: the memory API and proxy, the MCP server, and per-user vaults for your own product. See the honest comparisons for where the others fit better.
Planning a setup? Browse the integration docs, compare the coding-tool connection options, or follow the OpenClaw multi-agent guide. For existing files and conversations, start with the import overview.