home / mcp / echovault mcp server

EchoVault MCP Server

Local MCP server that stores, searches, and contextualizes coding memories on your machine.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "go-ports-echovault": {
      "command": "memory",
      "args": [
        "mcp"
      ]
    }
  }
}

EchoVault is a local MCP server that enables you to store, search, and retrieve coding memories on your machine. It exposes memory_save, memory_search, and memory_context as MCP tools so your agents can remember decisions, bugs, and context across sessions without cloud services or API keys.

How to use

You run the MCP server and connect your coding agents to it. The server offers three tools: memory_save to persist memories, memory_search to find relevant memories, and memory_context to load prior decisions and context at session start. Memories stay local in your vault and can be searched via keyword or semantic search if you enable embeddings.

How to install

Prerequisites: you need a Go toolchain to build from source or you can download a pre-built binary for your platform.

Install the pre-built binary by downloading the latest release for your platform and placing the binary somewhere on your PATH.

Build from source using the included Makefile. The process produces a memory binary you can place in your PATH.

# Build from source
git clone https://github.com/go-ports/echovault.git
cd echovault
make build          # produces ./bin/memory
sudo cp bin/memory /usr/local/bin/

# First run and setup
memory init
memory setup claude-code   # or: cursor, codex, opencode
```}]} ,{

If you want to customize where memories are stored or adjust embedding configuration, you can generate starter config files and set memory location per project or globally.

Configuration and usage notes

The MCP server runs locally and is started via a dedicated command. You can use a per-project setup to tailor configurations for different agents. The server stores memories as Markdown with YAML frontmatter in a local vault, and it provides fast full-text search plus optional semantic search through embeddings.

# Start the MCP server using the standard client command
memory mcp
```}]}]},{

MCP tools exposed by EchoVault

EchoVault exposes three MCP tools for your agents:

memory_save — Persist a memory with details, category, and tags.
memory_search — Retrieve memories by keyword or semantic query (if embeddings are configured).
memory_context — Load prior memories and context as a session starts.

Available tools

memory_save

Persist a memory with details, tags, and context so it can be retrieved later.

memory_search

Query memories using keywords or semantic vectors to surface relevant past decisions and learnings.

memory_context

Load prior memories and session context at the start of a new work session.