Working With Mcp

---
name: working-with-mcp
description: Use when user mentions MCPProxy/MCP tools (e.g., "check buildkite mcp", "use slack mcp") or when you need to discover or call tools through MCPProxy - immediately checks if mcp__MCPProxy__* tools are available, suggests /mcp reconnect if missing (MCPProxy MCP server not connected), explains when to use MCP tools vs HTTP API for debugging
---

# Working with MCPProxy and MCP Tools

## Overview

MCPProxy can be accessed two ways:
1. **MCP Tools** (`mcp__MCPProxy__*`) - When MCPProxy is configured as an MCP server in Claude Code
2. **HTTP API** (`curl http://localhost:8080/...`) - For debugging MCPProxy itself

**This skill is for #1.** For debugging MCPProxy itself, use HTTP API approach sparingly.

## FIRST: Detect MCPProxy Connection

**When user mentions MCPProxy or MCP tools:**
- "check buildkite mcp"
- "use slack mcp tools"
- Any MCPProxy tool request

**YOU MUST IMMEDIATELY:**

1. **Look at your available tools** (passive observation - don't run commands)
2. **Scan for tools starting with `mcp__MCPProxy__`**

**Examples of what you're looking for:**
```
mcp__MCPProxy__retrieve_tools
mcp__MCPProxy__call_tool
mcp__MCPProxy__upstream_servers
```

**Decision:**
- ✅ **See `mcp__MCPProxy__*` tools?** → MCPProxy is connected, use them
- ❌ **Don't see `mcp__MCPProxy__*` tools?** → MCPProxy MCP server not connected

**If tools are missing, IMMEDIATELY tell user:**

> "MCPProxy MCP tools aren't available in this session. The MCPProxy MCP server isn't connected.
>
> Try the `/mcp` command and select MCPProxy to reconnect. Once reconnected, you'll have access to the `mcp__MCPProxy__*` tools."

**DO NOT:**
- Try bash commands like `claude mcp list` or `mcpproxy list-tools`
- Try HTTP API with curl
- Make multiple attempts with different approaches
- Try to work around the issue

**Missing `mcp__MCPProxy__*` tools = MCPProxy MCP server not connected = Suggest `/mcp reconnect`. That's it.**

## Core Decision Tree (Passive Observation)

**This is passive observation - look at tools you already have, don't run commands.**

```
Look at your available tools (the tool list in this session)
  ↓
Do you see ANY tools starting with mcp__MCPProxy__*?
│
├─ YES → MCPProxy is connected
│         ├─ Discover: mcp__MCPProxy__retrieve_tools
│         └─ Call: mcp__MCPProxy__call_tool
│
└─ NO → MCPProxy MCP server not connected
          └─ Tell user: "Try /mcp command to reconnect"
```

**Key points:**
- This is OBSERVATION, not execution
- Don't run `claude mcp list` or any bash command
- Just scan your available tool list
- Missing tools = connection issue, not "tools don't exist"
- Never fall back to HTTP API

## When to Use This Skill

**Use when:**
- You need to discover what tools MCPProxy exposes
- You want to call a tool through MCPProxy
- MCPProxy is configured but you're not sure if MCP tools are available
- User mentions working with MCP servers or tools

**Do NOT use when:**
- User is asking about non-MCP topics
- The task doesn't involve MCP tools

## What You're Looking For

**When MCPProxy IS connected, your tool list includes:**
```
Available tools:
- Bash
- Read
- Write
- mcp__MCPProxy__retrieve_tools    ← Look for these
- mcp__MCPProxy__call_tool          ← Look for these
- mcp__MCPProxy__upstream_servers   ← Look for these
- mcp__ide__getDiagnostics
- mcp__ide__executeCode
```

**When MCPProxy is NOT connected:**
```
Available tools:
- Bash
- Read
- Write
- mcp__ide__getDiagnostics
- mcp__ide__executeCode
← No mcp__MCPProxy__* tools = Connection issue
```

**Key distinction:**
- `mcp__ide__*` = Different MCP server (IDE integration)
- `mcp__MCPProxy__*` = MCPProxy tools for upstream servers
- Only `mcp__MCPProxy__*` tools indicate MCPProxy connection

## Core Pattern: Discover Then Call

```typescript
// 1. Search for relevant tools
mcp__MCPProxy__retrieve_tools({
  query: "keywords describing what you need",
  limit: 10
});

// 2. Examine tool schemas in results
// Look at inputSchema to understand parameters

// 3. Call the tool
mcp__MCPProxy__call_tool({
  name: "server:tool-name",  // From search results
  args_json: JSON.stringify({  // MUST be JSON string
    param1: "value",
    param2: "value"
  })
});
```

**Critical:** `args_json` must be a JSON string (use `JSON.stringify()`), not a plain object.

## Available MCP Tools

- `mcp__MCPProxy__retrieve_tools` - Search for tools across all upstream servers
- `mcp__MCPProxy__call_tool` - Execute a discovered tool
- `mcp__MCPProxy__upstream_servers` - Manage server configuration
- `mcp__MCPProxy__list_registries` - List available MCP registries
- `mcp__MCPProxy__search_servers` - Find new servers in registries
- `mcp__MCPProxy__read_cache` - Read paginated results
- `mcp__MCPProxy__quarantine_security` - Manage quarantined servers

## Quick Reference

| Task | Check This First | Then Use |
|------|------------------|----------|
| Discover tools | Are `mcp__MCPProxy__*` in tool list? | If yes: `retrieve_tools`<br>If no: Suggest `/mcp` |
| Call a tool | Are `mcp__MCPProxy__*` in tool list? | If yes: `call_tool`<br>If no: Suggest `/mcp` |
| Debug MCPProxy | N/A | Use HTTP API sparingly |

## If MCP Tools Are Missing

When `mcp__MCPProxy__*` tools are not in your tool list:

**Tell the user:**
> "MCPProxy MCP tools aren't available in this session. Try the `/mcp` command and select MCPProxy to reconnect."

**Do NOT:**
- Fall back to HTTP API (that's for debugging MCPProxy)
- Assume the tools don't exist
- Try bash commands
- Make multiple curl attempts

**Reality:** The tools exist when MCP connection is established. Missing tools = connection issue.

## Top 3 Common Mistakes

### 1. Invoking MCP Tools as Bash Commands

```bash
# ❌ WRONG - MCP tools are not bash commands
mcp__MCPProxy__retrieve_tools --query "slack tools"
```

```typescript
// ✅ CORRECT - Call as MCP tool (function)
mcp__MCPProxy__retrieve_tools({
  query: "slack tools"
})
```

**How to tell the difference:**
- `mcp__server__tool` pattern → MCP tool (function call)
- `mcpproxy`, `docker` → Bash command (via Bash tool)

### 2. Using HTTP API Instead of MCP Tools

```bash
# ❌ WRONG - Don't do this when MCP tools should work
curl http://127.0.0.1:8080/api/v1/tools?apikey=...
```

```typescript
// ✅ CORRECT - Use MCP tools
mcp__MCPProxy__retrieve_tools({query: "search query"})
```

**HTTP API is only for debugging MCPProxy itself.**

### 3. Falling Into the HTTP API Spiral

```
Agent tries: mcp__MCPProxy__call_tool(...)
Gets error: "No such tool available"
Agent makes:
1. curl attempt with python parsing
2. curl attempt with json.tool
3. curl attempt with grep
4. Python script wrapping curl
5. Different parsing strategy...
```

**STOP!** Multiple curl attempts = you're avoiding the real problem: MCP not connected. Suggest `/mcp` instead.

## Red Flags - STOP Immediately

If you think any of these thoughts, **STOP**:

- "I'll run `mcp__MCPProxy__retrieve_tools` as a bash command"
- "The `mcp__MCPProxy__*` tools don't exist, I'll use curl"
- "HTTP API is simpler, I'll use that"
- "Let me try a different way to parse the HTTP response"
- "Maybe if I use Python/jq/grep it will work"
- "The tool error means MCPProxy isn't running"

**Reality:**
- `mcp__MCPProxy__*` tools exist when MCP connection is established
- Missing tools = connection issue, not "tools don't exist"
- HTTP API is for debugging MCPProxy, not for normal usage
- Running ≠ Connected. Process can run but MCP not connected.

### Red Flags in Error Responses

When an MCP tool call returns an error, check for these patterns. If matched, **STOP** - retrying won't help.

**Unrecoverable errors - STOP and tell the user:**

| Error Pattern | What It Means | Tell User |
|--------------|---------------|-----------|
| `authentication failed`, `OAuth/token authentication required`, `authorization required` | Upstream server needs re-auth | "Server '{name}' needs re-authentication. Run `mcpproxy auth login --server={name}` or use the MCPProxy system tray to re-authenticate." |
| `is not connected`, `is disabled` | Server offline/disabled | "Server '{name}' isn't connected. Check MCPProxy status with `mcp__MCPProxy__upstream_servers(operation='list')`." |
| `access_denied`, `insufficient_scope` | Missing permissions | "Server '{name}' lacks permissions for this operation. May need to re-authorize with additional scopes." |

**Why retrying won't help:**

These errors require user action outside of Claude:
- Re-running OAuth flow through MCPProxy CLI or system tray
- Fixing server configuration
- Granting additional permissions

**Don't:**
- Retry the same tool with different parameters
- Try other tools on the same server (they'll fail too)
- Fall back to WebFetch/curl (won't have auth either)

## Debugging MCPProxy Connection Issues

If MCPProxy itself isn't working properly (servers won't connect, Docker issues, etc.), you can use these quick checks:

### Quick Health Check

```bash
# 1. Is mcpproxy running?
ps aux | grep mcpproxy | grep -v grep

# 2. Get API key
grep '"api_key"' ~/.mcpproxy/mcp_config.json

# 3. Check server status
curl -s "http://127.0.0.1:8080/api/v1/servers?apikey=YOUR_KEY" | python3 -m json.tool

# 4. Check for recent errors
tail -50 ~/Library/Logs/mcpproxy/main.log | grep -i error
```

**Key status fields to check:**
- `connected`: Boolean - is the server connected?
- `status`: String - current state (connecting, ready, error)
- `last_error`: String - most recent error message
- `tool_count`: Number - how many tools available

### Common MCPProxy Issues

#### "the input device is not a TTY" (Docker servers)
**Fix:** Add `"isolation": {"enabled": false}` to the Docker-based server config.

#### "unexpected argument found" (uvx/npx servers)
**Fix:** Put package name as first arg: `["mcp-server-name", "--arg", "value"]`

#### "Invalid or missing API key" after restart
**Fix:** Check `echo $MCPPROXY_API_KEY` - environment variable overrides config file.

### Restart MCPProxy

```bash
pkill mcpproxy
sleep 2
open /Applications/mcpproxy.app  # macOS
# OR
mcpproxy &  # Linux/headless
```

## Summary

**The capability:** Use MCPProxy's tool discovery to find and call tools across all upstream MCP servers.

**The pattern:**
1. Check if `mcp__MCPProxy__*` tools exist
2. If yes → Use them (discover then call)
3. If no → Suggest `/mcp` reconnect
4. Never fall back to HTTP API

**Result:** Fast tool usage with clear error handling.