# CoralSwarm Connect

This MCP server stores coding context, decisions, and sessions in CoralSwarm, then lets an agent retrieve them. It fits work that continues across sessions.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI
- Author: CoralSwarm
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/CoralSwarm/coralswarm-connect
- Updated: 2026-10-11
- HTML: https://affpapa.org/en/nevibecode/mcp/coralswarm-connect

## What it does

CoralSwarm Connect exposes CoralSwarm tools such as `list_oceans`, `recent_activity`, `ask_ocean`, `search_atoms`, `get_reef`, and `add_context`. An agent can search saved material, choose a personal or organization workspace, and write new context to it.

The plugin declares the MCP server, bundles a skill, and runs hooks for `SessionStart`, `UserPromptSubmit`, `PreCompact`, `Stop`, and `PostToolUse`. These hooks prompt mid-session saves, flush pending work before compaction, and record `add_context` activity for recovery. Sign-in uses OAuth through `/mcp`; there is no API key to paste.

Use the exact server URL `https://api.coralswarm.com/mcp`, since Cursor rejects a resource with a different protected-resource address. Enabling two plugin copies causes `SessionStart` to fire twice. That is a configuration conflict worth avoiding.

## Requirements
- Remote server: `https://api.coralswarm.com/mcp`

## Install

### Claude Code (native platform)

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

### Codex

**Platform CLI: remote server**
```bash
codex mcp add coralswarm-connect --url https://api.coralswarm.com/mcp
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "coralswarm-connect": {
            "url": "https://api.coralswarm.com/mcp"
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "coralswarm-connect": {
            "type": "http",
            "url": "https://api.coralswarm.com/mcp"
        }
    }
}
```

### Gemini CLI

**Platform CLI: remote server**
```bash
gemini mcp add -t http coralswarm-connect https://api.coralswarm.com/mcp
```

## When to use it
- save session decisions to CoralSwarm
- search earlier project context
- switch between personal and organization workspaces
- recover context after a crashed session

## Example prompts
- Save this milestone with `add_context` and include the current Git branch.
- Search CoralSwarm for prior decisions related to this error with `search_atoms`.
- Call `list_tenants`, then use `select_tenant` for the organization workspace.

## Similar
- [vercel-connect](https://affpapa.org/en/nevibecode/skills/vercel-connect.md) — skill · 1 install
- [sumsub-integrate-id-connect](https://affpapa.org/en/nevibecode/skills/sumsub-integrate-id-connect.md) — skill · 980 installs
- [rekindle](https://affpapa.org/en/nevibecode/mcp/rekindle.md) — MCP server · ★ 5
- [desk-journal](https://affpapa.org/en/nevibecode/skills/desk-journal.md) — skill · 25 installs
- [agent-memory](https://affpapa.org/en/nevibecode/plugins/agent-memory.md) — plugin · ★ 28854
- [agentmemory-hooks](https://affpapa.org/en/nevibecode/skills/agentmemory-hooks.md) — skill · 10800 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