Memori MCP

With Memori's MCP server, your agent can retrieve relevant memories before answering and store durable facts after responding, keeping context across sessions without any SDK integration.

Memori MCP

Persistent AI memory for any MCP-compatible agent — no SDK required.

memori-mcp is the official Memori MCP server. Connect it to your AI agent to give it long-term memory: recall relevant facts before answering, store durable preferences after responding, and maintain context across sessions.


Why Memori MCP?

Memori turns stateless agents into stateful systems by providing structured, persistent memory that works across sessions and workflows.

  1. Persistent state beyond prompts — Most agents rely on prompt context and lose state between runs. Memori provides durable, structured memory so agents can retain facts, decisions, and outcomes over time.
  2. Memory from execution (not just natural language) — Traditional systems extract memory from chat. Memori builds memory from agent execution itself — including tool calls, decisions, and results. This enables true agent-native memory, not just conversational recall.
  3. Lower cost, higher accuracy — Instead of expanding prompt context, Memori retrieves only what matters.
    • Significantly reduced token usage
    • Faster responses
    • Improved accuracy vs long-context approaches
  4. Works with any MCP client and production-ready - No SDK, no code changes, just config

Memori is state infrastructure for production agents — enabling persistent memory, efficient retrieval, and structured context across both natural language and agent execution.

LoCoMo Benchmark

Memori was evaluated on the LoCoMo benchmark for long-conversation memory and achieved 81.95% overall accuracy while using an average of 1,294 tokens per query. That is just 4.97% of the full-context footprint, showing that structured memory can preserve reasoning quality without forcing large prompts into every request.

Compared with other retrieval-based memory systems, Memori outperformed Zep, LangMem, and Mem0 while reducing prompt size by roughly 67% vs. Zep and lowering context cost by more than 20x vs. full-context prompting.

Read the benchmark overview or download the paper.


How It Works

The server exposes two tools:

ToolWhen to callWhat it does
recallStart of each user turnFetches relevant memories for the current query
advanced_augmentationAfter composing a responseStores durable facts and preferences for future sessions

Example Agent Flow

Given the message: "I prefer Python and use uv for dependency management."

  1. Agent calls recall with the user message as query
  2. Agent uses any returned facts to compose a response
  3. Agent calls advanced_augmentation with the user message and response

On a later turn — "Write a hello world script" — the agent recalls the Python + uv preference and personalizes its response automatically.


Prerequisites

  • A Memori API key from app.memorilabs.ai
  • An entity_id to identify the end user (e.g. user_123)
  • An optional process_id to identify the agent or workflow (e.g. my_agent)

Export these in your shell or replace the placeholders directly in your config:

export MEMORI_API_KEY="your-memori-api-key"
export MEMORI_ENTITY_ID="user_123"
export MEMORI_PROCESS_ID="my_agent"   # optional

Server Details

PropertyValue
Endpointhttps://api.memorilabs.ai/mcp/
TransportStateless HTTP
AuthAPI key via request headers

Headers

HeaderRequiredDescription
X-Memori-API-KeyYesYour Memori API key
X-Memori-Entity-IdYesStable end-user identifier (e.g. user_123)
X-Memori-Process-IdNoProcess, app, or workflow identifier for memory isolation

session_id is derived automatically as <entity_id>-<UTC year-month-day:hour> — you do not need to provide it.


Verifying the Connection

After configuring any client:

  1. Confirm the MCP server shows as connected in your client's UI
  2. Check that recall and advanced_augmentation appear in the tools list
  3. Send a test message — recall should return a response (even if empty for new entities)
  4. Verify advanced_augmentation returns memory being created

If you receive 401 errors, double-check your X-Memori-API-Key value. See the Troubleshooting guide for more help.


Links

関連サーバー