# dakera-mcp

A self-hosted MCP server that gives agents persistent, searchable memory. Use it when context must carry across sessions.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 9
- Author: Dakera-AI
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/Dakera-AI/dakera-mcp
- Updated: 2026-10-22
- HTML: https://affpapa.org/en/nevibecode/mcp/dakera-mcp

## What it does

dakera-mcp gives an MCP agent persistent memory. It can save entries, retrieve them by meaning, search with tag and type filters, manage sessions, build a knowledge graph, and extract entities from text. Hybrid search combines vector retrieval with BM25.

The default profile exposes 14 tools. An agent can inspect the wider catalog with `dakera_discover_tools` and load selected schemas through `dakera_load_tools`; profiles extend the set to 106 tools. The server is distributed as a single Rust binary and runs through Docker.

\`dakera-mcp 0.12.3` works with Dakera servers `v0.11.108` and `v0.12.0`–`v0.12.3`. Docker is required for the installation command to start the server.

## Requirements
- Needs: Docker

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add dakera-mcp -- docker run -i --rm ghcr.io/dakera-ai/dakera-mcp:0.12.3
```

### Codex

**Platform CLI**
```bash
codex mcp add dakera-mcp -- docker run -i --rm ghcr.io/dakera-ai/dakera-mcp:0.12.3
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "dakera-mcp": {
            "command": "docker",
            "args": [
                "run",
                "-i",
                "--rm",
                "ghcr.io/dakera-ai/dakera-mcp:0.12.3"
            ]
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "dakera-mcp": {
            "type": "stdio",
            "command": "docker",
            "args": [
                "run",
                "-i",
                "--rm",
                "ghcr.io/dakera-ai/dakera-mcp:0.12.3"
            ]
        }
    }
}
```

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add dakera-mcp docker run -i --rm ghcr.io/dakera-ai/dakera-mcp:0.12.3
```

## When to use it
- carry decisions and facts across agent sessions
- retrieve stored context from a natural-language query
- filter memories by tags, types, or time
- build a knowledge graph from stored data
- remove one memory or a matching group

## Example prompts
- Save the architecture decision from this conversation with the tag `architecture`.
- Find earlier decisions about authentication and show the related memories.
- End the current session and attach a short summary.

## Similar
- [agent-memory-systems](https://affpapa.org/en/nevibecode/skills/agent-memory-systems.md) — skill · 1086 installs
- [honcho-integration](https://affpapa.org/en/nevibecode/skills/honcho-integration.md) — skill · 883 installs
- [neo4j-agent-memory-skill](https://affpapa.org/en/nevibecode/skills/neo4j-agent-memory-skill.md) — skill · 879 installs
- [Honcho](https://affpapa.org/en/nevibecode/mcp/honcho.md) — MCP server · ★ 7784
- [ResiMind](https://affpapa.org/en/nevibecode/mcp/resimind.md) — MCP server · ★ 292
- [ContextLattice](https://affpapa.org/en/nevibecode/mcp/contextlattice.md) — MCP server · ★ 156

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md