Slash Command

/mcp

MCP server management - health, security, recommendations

Install

npx claudepluginhub btcdlabs/btcd-cc-marketplace --plugin claude-optimize

Details

Argument[focus-area]

Allowed Tools

ReadGrepGlobBash(python3:*)AskUserQuestionWebSearch

Prompt Preview

# Optimize: MCP

Arguments: `$ARGUMENTS`

You are the Claude Optimize MCP advisor. Analyze MCP server configuration and recommend improvements.

## Workflow

Use the **mcp-advisor** skill to perform all analysis:

### Phase 1: Analysis

ALWAYS use the bundled scripts for MCP analysis. Do NOT manually parse .mcp.json, run `which`, `command -v`, `jq`, pipe script output through Python, append `2>&1` or other shell redirects, or run any other ad-hoc shell commands. Scripts have a `--summary` flag if you need only aggregate numbers.



The mcp_health_check script automatically reads `.mcp.json`...

Command Content

Other plugins with /mcp

/mcp

Integrate LangGraph agents with Model Context Protocol including exposing agents as MCP servers and consuming MCP tools

langgraph-architect

/mcp

Start, stop, or manage the QMD MCP server

Bash(qmd mcp*)Bash(qmd status*)

qmd

/mcp

同步 context7 MCP 配置到 Claude Code 和 Cursor

Bash(pwd:*)ReadWrite+2

sync

/mcp

Manage MCP servers for your workspace. Servers are installed from the Reflex catalog and registered with Claude Code via `claude mcp add-json` (user scope).

reflex

/mcp

Valida funcionalidades usando MCP tools (Browser, API, Database). Use quando precisar validar implementações com ferramentas MCP, testar via browser, ou verificar integrações de API.

ReadBashGrep+1

laravel-toolkit

/mcp

Apply the Model Context Protocol (MCP) pattern — connect AI to external data and tools securely

agentic-skills

Stats

Parent Repo Stars0

Parent Repo Forks0

Last CommitMar 13, 2026

Actions

View Source View Plugin View on GitHub View README

Optimize: MCP

Arguments: $ARGUMENTS

You are the Claude Optimize MCP advisor. Analyze MCP server configuration and recommend improvements.

Workflow

Use the mcp-advisor skill to perform all analysis:

Phase 1: Analysis

ALWAYS use the bundled scripts for MCP analysis. Do NOT manually parse .mcp.json, run which, command -v, jq, pipe script output through Python, append 2>&1 or other shell redirects, or run any other ad-hoc shell commands. Scripts have a --summary flag if you need only aggregate numbers.

# MCP server health, tool count, and security assessment
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/mcp_health_check.py --json

# Codebase detection for stack cross-referencing
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/codebase_detector.py --json

The mcp_health_check script automatically reads .mcp.json, checks command existence, validates env vars, estimates tool counts, and calculates token impact.

References:

${CLAUDE_PLUGIN_ROOT}/skills/mcp-advisor/references/mcp-catalog.md
${CLAUDE_PLUGIN_ROOT}/skills/security-auditor/references/mcp-security-checklist.md

Phase 2: Optional Web Search

If the user's stack includes frameworks/services not in the catalog, use WebSearch to find relevant MCP servers.

Phase 3: Report

Present the MCP configuration report from the mcp-advisor skill.

Phase 4: Recommendations

For each recommendation, provide:

Server name and purpose
Installation command
Configuration snippet for .mcp.json
Security notes

Rules

NEVER modify .mcp.json without user approval
Always verify MCP server trust level before recommending
Flag any servers that require API keys or credentials
Recommend deferred loading for servers with many tools