# MCP Manager Skill

**Intelligent on-demand MCP server loading to minimize context consumption**

## Quick Start

The skill operates automatically - Jarvis will detect when you need MCPs and offer to load them.

### Example Workflow

**You:** "Check what tech stack reddit.com is using"

**Jarvis:**
```
🔍 Detected need for: httpx
Loading MCP for web stack analysis...

✅ Enabled httpx (~300 tokens)
⚠️  Please restart Claude Code to activate, then resubmit your request.
```

**After restart:**
```
[Jarvis completes the task using httpx MCP]

Task complete!
Currently loaded MCPs: httpx (~300 tokens)
Would you like me to unload this to free context? (yes/no)
```

## Manual Commands

You can also manually control MCPs:

```bash
# Check what's currently loaded
python3 ~/.claude/skills/mcp-manager/mcp_control.py status

# Load specific MCP
python3 ~/.claude/skills/mcp-manager/mcp_control.py enable httpx

# Unload specific MCP
python3 ~/.claude/skills/mcp-manager/mcp_control.py disable httpx

# See what MCPs would help with a task
python3 ~/.claude/skills/mcp-manager/mcp_control.py detect "your query here"

# Reset to minimal (unload all)
python3 ~/.claude/skills/mcp-manager/mcp_control.py reset
```

Or just ask Jarvis:
- "Load the httpx MCP"
- "Unload all MCPs"
- "What MCPs are currently loaded?"
- "Show MCP status"

## Available MCPs

### Security & Analysis
- **httpx** - Web server tech stack analysis (~300 tokens)
- **naabu** - Port scanning and service discovery (~300 tokens)

### Web Scraping
- **apify** - Web scraping via Apify platform (~800 tokens)
- **brightdata** - Enterprise data extraction (~500 tokens)
- **playwright** - Browser automation (~1000 tokens)

### Development
- **Ref** - Documentation search (~600 tokens)
- **stripe** - Payment processing (~700 tokens)

### Personal (Daniel Miessler)
- **content** - Content archive (~400 tokens)
- **daemon** - Personal API (~300 tokens)
- **Foundry** - PAI infrastructure (~500 tokens)

## Context Savings

- **All MCPs loaded**: ~5400 tokens
- **Minimal (none loaded)**: ~0 tokens
- **Savings per session**: 5000+ tokens available for actual work!

## How Detection Works

The skill scans your request for trigger keywords:

- "tech stack", "what technology" → `httpx`
- "port scan", "open ports" → `naabu`
- "scrape", "crawl website" → `apify`, `brightdata`
- "browser automation" → `playwright`
- "documentation", "API docs" → `Ref`
- "stripe", "payment" → `stripe`

## Files

- **SKILL.md** - Main skill documentation (loaded by Jarvis)
- **mcp_control.py** - MCP management CLI tool
- **~/.claude/.mcp.json** - Active MCP configuration
- **~/.claude/.mcp-full.json** - Backup of all MCPs
- **~/.claude/.mcp.json.backup** - Previous config backup
- **~/.claude/logs/mcp-manager.log** - Change log

## Important Notes

⚠️ **Restart Required**: MCP changes require Claude Code restart to take effect

🔒 **Credentials Safe**: The tool preserves all API keys and auth tokens

💾 **Auto-Backup**: Configurations are backed up before every change

📊 **Token Estimates**: Token counts are approximations based on MCP complexity

## Workflow

1. **Minimal Start** - Begin each session with no MCPs loaded
2. **Auto-Detect** - Jarvis detects when MCPs are needed
3. **Load on Demand** - Enable only required MCPs
4. **Complete Task** - Use the MCP for your work
5. **Unload** - Free context by removing MCPs when done
6. **Repeat** - Next task loads different MCPs as needed

This approach keeps your context budget focused on your actual work instead of unused MCP definitions.

---

**Version**: 1.0.0
**Created**: 2025-11-09
**Maintained by**: Jarvis (PAI System)