# sage

A local MCP server that stores agent experience, checks new records, and recalls context in later sessions. Built for work that continues over time.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 257
- Author: l33tdawg
- License: Apache-2.0 · policy=link-only · description only; the file stays with the author
- Source: https://github.com/l33tdawg/sage
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/mcp/sage

## What it does

\`sage` connects an MCP client to a local memory node. An agent can save design choices, failed approaches, and lessons for later sessions, then retrieve context related to the current work. New records go through an admission check, carry an agent identity and access rules, and can be inspected in the local `CEREBRUM` dashboard.

Memory stays local by default. A personal setup uses one validator, while a multi-validator network relies on a BFT quorum. These checks decide whether a record is accepted; they do not establish that every remembered claim is true or alter model weights. Recall depends on recorded content, configuration, and whether the agent uses its tools.

The server connects to Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, and Zed. The command-line setup needs Docker because it starts `ghcr.io/l33tdawg/sage:11.24.2`. Developers can also integrate through the Python SDK or a signed REST API.

## Requirements
- Needs: Docker

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add sage -- docker run -i --rm ghcr.io/l33tdawg/sage:11.24.2
```

### Codex

**Platform CLI**
```bash
codex mcp add sage -- docker run -i --rm ghcr.io/l33tdawg/sage:11.24.2
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "sage": {
            "command": "docker",
            "args": [
                "run",
                "-i",
                "--rm",
                "ghcr.io/l33tdawg/sage:11.24.2"
            ]
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "sage": {
            "type": "stdio",
            "command": "docker",
            "args": [
                "run",
                "-i",
                "--rm",
                "ghcr.io/l33tdawg/sage:11.24.2"
            ]
        }
    }
}
```

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add sage docker run -i --rm ghcr.io/l33tdawg/sage:11.24.2
```

## When to use it
- carry decisions across agent sessions
- retrieve earlier context for a task
- control which agents can use a memory
- trace who created a record
- review the history of stored experience

## Example prompts
- Save the decision about this library so it is available in later sessions.
- Recall which approach to this task already failed.
- Show me in `CEREBRUM` what the agent stored for this project.

## Similar
- [REA](https://affpapa.org/en/nevibecode/mcp/rea.md) — MCP server · ★ 386989
- [GitHub](https://affpapa.org/en/nevibecode/mcp/github.md) — MCP server · ★ 33918
- [World Monitor](https://affpapa.org/en/nevibecode/mcp/world-monitor.md) — MCP server · ★ 89408
- [Netdata](https://affpapa.org/en/nevibecode/mcp/netdata.md) — MCP server · ★ 81115
- [Front-End Checklist](https://affpapa.org/en/nevibecode/mcp/front-end-checklist.md) — MCP server · ★ 74559
- [Scrapling MCP Server](https://affpapa.org/en/nevibecode/mcp/scrapling-mcp-server.md) — MCP server · ★ 89433

---
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