Integrations

Connect your chat assistants, coding agents, frameworks, and existing history to a MemoryRouter vault.

MemoryRouter adds persistent conversational memory across sessions. Choose the guide for the tool you actually use, then connect the same vault wherever you want that context available.

Create a MemoryRouter account or compare all integrations.

Start with your primary tool

Integration	Setup guide
Claude Code	Install memory hooks for the Claude Code CLI.
Codex	Set up Codex hooks and memory tools.
ChatGPT	Connect MemoryRouter through ChatGPT Developer mode.
Claude Cowork	Connect Cowork to your selected vault.
Claude	Add a custom connector in Claude.

Native hooks and model-directed connectors behave differently. Claude Code and Codex have hook integrations. ChatGPT, Claude, and Cowork use tools that the model decides when to call. See the platform comparison before choosing a setup.

Chat assistants

Open WebUI: connect a self-hosted, multi-model chat interface through a Filter Function.
ChatGPT, Claude Cowork, and Claude connectors: the primary chat guides above.

Coding agents

Integration	Setup guide
OpenClaw	Install the mr-memory plugin.
DeepSeek Harness	Connect DeepSeek Harness with dsh-memoryrouter.
Hermes	Install the Hermes memory provider.
NemoClaw	Connect the OpenClaw-compatible agent inside its sandbox.
Cursor	Configure memory tools in Cursor.
Windsurf	Configure memory tools in Windsurf.
Cline	Configure memory tools in Cline.
Continue	Configure memory tools in Continue.
Aider	Use the supported provider proxy path in Aider.
Cody	Check the supported configuration path and limits.
Zed	Configure memory tools in Zed.

The coding tools guide distinguishes native integrations, MCP tools, and provider configuration. A generic MCP connection does not add automatic capture hooks.

Frameworks and protocols

Integration	Setup guide
MCP	Remote endpoint, authentication, scopes, and memory tools.
LangChain	Use memory tools in a Python agent.
LangGraph	Add recall and retain nodes to a graph.
Proxy API	Route supported model requests through MemoryRouter.
CLI	Install the MemoryRouter command-line tools.
Local inference	Keep your model call and use prepare / ingest.
Supported models	Choose a provider path and check model compatibility.

Data and imports

Integration	Setup guide
Memory transfer	Review and carry useful context between AI tools.
ChatGPT archive import	Import a selected official ChatGPT export.
Agent history import	Load existing agent transcripts and workspace context.
Vault transfer	Copy approved memory between MemoryRouter vaults.

For custom JSONL and application-owned records, start with imports. Connecting a tool does not itself import that tool's old conversations.

Verify your connection

Use a synthetic fact, not personal or customer information:

In session A, ask your connected tool to save: For the demo project, the review day is Thursday and the label is harbor-copper-42.
Confirm the tool reports an actual save. For hook-based integrations, follow that guide's capture and flush instructions.
Open a new session with the same vault and ask: Recall the demo project's review day and label from MemoryRouter. Do not include the expected answer in the question.
Check the recalled result against session A. If it is missing, use that integration's troubleshooting section instead of assuming the connection worked.

This is a test recipe, not a claim that a test was run in your account. See security and data isolation before connecting real data.