# Willder

The Willder MCP server for Claude Code shares memory across AI agents and gives each one scoped, expiring access. Agents receive task context and leave an audit trail when they change or read facts.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 12 repo stars
- Author: willderai
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/willderai/willder-core
- Updated: 2026-11-04
- Security check: automated, 2026-11-17 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/mcp/willder

## What it does

Willder lets one agent save a fact for another to retrieve on a later request. Search can match wording, meaning, and links between facts. Each agent has a token tied to a scope and an expiry; revoking it blocks that agent’s next call. Records retain authorship and change history.

GitHub, Google Drive, Slack, Figma, and PostHog can sync into a chosen scope. A self-hosted setup needs Docker with Compose v2, Node 20.19+, 22.12+, or 24+, plus several gigabytes of disk space. Set `GRAPHITI_SERVICE_API_KEY` in the configuration; any string works. The hosted endpoint is `https://willder.ai/api/mcp`; connecting to a local MCP server uses a separate setup.

## Requirements
- Remote server: `https://willder.ai/api/mcp`

## Install

### Claude Code (native platform)

**Platform CLI: remote server**
```bash
claude mcp add --transport http willder https://willder.ai/api/mcp
```

### Codex

**Platform CLI: remote server**
```bash
codex mcp add willder --url https://willder.ai/api/mcp
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "willder": {
            "url": "https://willder.ai/api/mcp"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "willder": {
            "type": "http",
            "url": "https://willder.ai/api/mcp"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http willder https://willder.ai/api/mcp
```

## When to use it
- share a saved fact between separate agents
- limit an agent’s memory access to a scope
- revoke a token to block its next request
- trace which agents wrote or read a fact

## Example prompts
- Save a note in Willder that the test environment uses Postgres, then retrieve it.
- Give Codex the API structure decision saved by Claude Code.
- Show which agent recorded the launch date and which agents read that fact.

## Similar
- [Filelayer](https://affpapa.org/en/nevibecode/mcp/filelayer.md) — MCP server
- [sanctuary](https://affpapa.org/en/nevibecode/mcp/sanctuary.md) — MCP server · ★ 10 repo stars
- [neo4j-security-skill](https://affpapa.org/en/nevibecode/skills/neo4j-security-skill.md) — skill · 874 installs
- [januscope](https://affpapa.org/en/nevibecode/mcp/januscope.md) — MCP server · ★ 33 repo stars
- [SchemaBrain](https://affpapa.org/en/nevibecode/mcp/schemabrain.md) — MCP server · ★ 12 repo stars
- [neo4j-agent-memory-skill](https://affpapa.org/en/nevibecode/skills/neo4j-agent-memory-skill.md) — skill · 940 installs

---
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
