# Ainative Zerodb Memory MCP

> Ainative Zerodb Memory MCP is an MCP server for knowledge and memory. Persistent semantic memory for AI agents on ZeroDB, with context pruning. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Platforms:** CLI, API, self-hosted
- **Repository:** https://github.com/ainative-studio/ainative-zerodb-memory-mcp
- **MCP Registry name:** io.github.AINative-Studio/ainative-zerodb-memory-mcp
- **Categories:** [knowledge and memory](https://ai-powered.com/categories/knowledge-and-memory), [communication](https://ai-powered.com/categories/communication), [AI models and LLM tooling](https://ai-powered.com/categories/ai-and-llm)
- **Listing updated:** 2026-11-15

## About

This is the slimmed-down memory server from AINative's ZeroDB. The README says it replaced a monolithic 77-tool server that used over 10,400 tokens of context with a focused set of about 1,400 tokens, an 87% reduction; the registry entry counts 18 tools including write-back to Slack, Gmail, Calendar and GitHub.

Memories are stored with automatic importance scoring (0.0 to 1.0) and embeddings from BAAI BGE models in 384, 768 or 1024 dimensions. Retrieval is semantic, can search across sessions with an agent scope, and can filter by importance, tags and similarity. A context tool prunes to a token limit while keeping important and recent items and optionally applying decay, and a synthesis tool asks Claude Haiku to turn relevant memories into a narrative, bullet or structured summary with a fallback to concatenated snippets.

It runs over stdio against either a local ZeroLocal endpoint on localhost:8000 or the managed cloud API at api.ainative.studio, authenticating with an API key (preferred) or username and password plus a project ID.

## Features

- **Auto-scored memories:** Stores conversation context with importance scoring and automatic embeddings.
- **Semantic search:** Finds memories by meaning, across sessions, with importance and similarity filters.
- **Token-limited context:** Prunes a session's memories to fit a token budget, keeping important and recent ones.
- **Context synthesis:** Summarises retrieved memories as narrative, bullet or structured text.
- **Local or cloud:** Works against ZeroLocal on localhost or the ZeroDB cloud API, auto-detecting the endpoint.
- **Session clearing:** Deletes a session's memories after confirmation, optionally keeping important ones.

## Good for

- Give a coding agent memory that persists between sessions
- Recall a user's stated preferences in a later conversation
- Keep prompt context under a token limit without losing key facts
- Summarise past decisions on a topic for a new session

## Install

### Claude Code (Terminal)

```sh
claude mcp add ainative-zerodb-memory-mcp -- npx -y ainative-zerodb-memory-mcp
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "ainative-zerodb-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "ainative-zerodb-memory-mcp"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "ainative-zerodb-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "ainative-zerodb-memory-mcp"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "ainative-zerodb-memory-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "ainative-zerodb-memory-mcp"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "ainative-zerodb-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "ainative-zerodb-memory-mcp"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "ainative-zerodb-memory-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "ainative-zerodb-memory-mcp"
      ]
    }
  }
}
```

## Tools

- `zerodb_store_memory`: Stores context with automatic importance scoring and embedding.
- `zerodb_search_memory`: Searches memory semantically with natural language.
- `zerodb_get_context`: Returns session context pruned to a token limit.
- `zerodb_embed_text`: Generates vector embeddings for text.
- `zerodb_semantic_search`: Searches by vector similarity without a text query.
- `zerodb_clear_session`: Clears a session's memories, optionally keeping important ones.
- `zerodb_synthesize_context`: Retrieves and synthesises memories into a coherent context string.
- `zerodb_configure_auto_context`: Enables auto-context middleware.

## FAQ

### How do I authenticate?

Set ZERODB_API_KEY, ZERODB_API_URL and ZERODB_PROJECT_ID. The README prefers the API key over username and password.

### Can it run fully locally?

Yes. It can point at ZeroLocal on localhost:8000, or at the cloud API, or both as two servers.

### Which embedding models are used?

BAAI BGE models in 384, 768 and 1024 dimensions.

### How many tools does it have?

The README says 14, while the registry entry says 18 including write-back actions.

### How do I add Ainative Zerodb Memory MCP to Claude Code?

Run this command in your terminal: claude mcp add ainative-zerodb-memory-mcp -- npx -y ainative-zerodb-memory-mcp

### What are the best alternatives to Ainative Zerodb Memory MCP?

The closest alternatives to Ainative Zerodb Memory MCP are VelesDB Memory, Neuromcp, Longterm Memory, ZenBrain Memory and HeurChain — AI Agent Memory.

## Where this comes from

- Description: written from the project's README.
- Install steps: from the official MCP Registry entry.

## Alternatives

- [VelesDB Memory](https://ai-powered.com/mcp/velesdb-memory): VelesDB Memory is an MCP server for knowledge and memory.
- [Neuromcp](https://ai-powered.com/mcp/neuromcp): Neuromcp is an open-source MCP server for knowledge and memory.
- [Longterm Memory](https://ai-powered.com/mcp/longterm-memory): Longterm Memory is an open-source MCP server for knowledge and memory.
- [ZenBrain Memory](https://ai-powered.com/mcp/zenbrain-memory): ZenBrain Memory is an open-source MCP server for knowledge and memory.
- [HeurChain — AI Agent Memory](https://ai-powered.com/mcp/heurchain-ai-agent-memory): HeurChain — AI Agent Memory is an MCP server for knowledge and memory.
- [Hebby Memories](https://ai-powered.com/mcp/hebby-memories): Hebby Memories is a paid MCP server for knowledge and memory.
- [agentmem](https://ai-powered.com/mcp/agentmem): agentmem is an open-source MCP server for knowledge and memory.
- [Komplyt Zero](https://ai-powered.com/mcp/komplyt-zero): Komplyt Zero is an open-source MCP server for knowledge and memory.

---
Source: https://ai-powered.com/mcp/ainative-zerodb-memory-mcp
