# phantom-ai-coworker

phantom-ai-coworker is a Claude Code skill that gives an agent its own machine, vector memory, Slack/email channels, and an MCP server that creates tools at runtime.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode, Windsurf (Devin Desktop), Claude.ai · 781 installs · ★ 98 repo stars
- Author: reason-machines
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/reason-machines/trending-skills/tree/HEAD/skills/phantom-ai-coworker/SKILL.md
- Updated: 2026-10-09
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/skills/phantom-ai-coworker

## What it does

The agent runs on a separate VM with Docker, stores session context in Qdrant, registers new MCP tools at runtime, and communicates via Slack, Telegram or webhook. Built on Claude Agent SDK with TypeScript/Bun, vector embeddings via Ollama, a self-evolution engine (observe → reflect → propose → validate), and an AES-256-GCM encrypted credential vault.

Requires `ANTHROPIC_API_KEY`, Qdrant running on `localhost:6333`, and Ollama for embeddings. Slack channel needs three tokens: `SLACK_BOT_TOKEN`, `SLACK_APP_TOKEN`, and `SLACK_SIGNING_SECRET`. The agent starts an HTTP server on port 3100 and serves an MCP endpoint at `http://localhost:3100/mcp`, which you can connect to Claude Code via `claude_desktop_config.json`.

The skill teaches the agent to set up Phantom: clone the repository, edit `.env`, run `docker compose up -d`, and check status via `/health`. Tools that Phantom registers dynamically will become available to all agents connected to its MCP server.

## Requirements
- Works with: Slack

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/trending-skills --skill phantom-ai-coworker -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/phantom-ai-coworker/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/phantom-ai-coworker/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/trending-skills --skill phantom-ai-coworker -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/phantom-ai-coworker/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/trending-skills --skill phantom-ai-coworker -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/phantom-ai-coworker/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install reason-machines/trending-skills skills/phantom-ai-coworker
```

2. **Manually: put the files in place**
- File: `.github/skills/phantom-ai-coworker/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add reason-machines/trending-skills --skill phantom-ai-coworker -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/reason-machines/trending-skills --path skills/phantom-ai-coworker
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/phantom-ai-coworker/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add reason-machines/trending-skills --skill phantom-ai-coworker -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/reason-machines/trending-skills/tree/HEAD/skills/phantom-ai-coworker/SKILL.md

## When to use it
- run an AI agent with persistent memory across sessions
- create MCP tools dynamically without editing config files
- give an agent access to Slack, email, and webhooks for autonomous operation
- deploy infrastructure (databases, dashboards, APIs) on a dedicated VM
- store project context in a vector database with semantic search

## Example prompts
- Set up a Phantom agent on Docker with Qdrant and a Slack channel
- Connect Phantom's MCP server to Claude Code at `http://localhost:3100/mcp`
- Build a new tool for Phantom that sends a report to Slack every morning

## Similar
- [setup-benny](https://affpapa.org/en/nevibecode/skills/setup-benny.md) — skill · 830 installs
- [oncall-irm](https://affpapa.org/en/nevibecode/skills/oncall-irm.md) — skill · 4600 installs
- [sentry-create-alert](https://affpapa.org/en/nevibecode/skills/sentry-create-alert.md) — skill · 4400 installs
- [authoring-error-tracking-alerts](https://affpapa.org/en/nevibecode/skills/authoring-error-tracking-alerts.md) — skill · 209 installs
- [composio](https://affpapa.org/en/nevibecode/skills/starchild-ai-agent-composio.md) — skill · 4300 installs
- [n8n-agents-official](https://affpapa.org/en/nevibecode/skills/n8n-agents-official.md) — skill · 2500 installs

From the same set: [reason-machines/trending-skills](https://affpapa.org/en/nevibecode/by/reason-machines/trending-skills.md) — 211 cards

---
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
