Skill

mcp-advisor

Use when running /optimize:mcp to analyze MCP server configuration, recommend new servers based on detected project stack, perform health checks, and assess security. Triggers on MCP server optimization, MCP recommendations, or MCP health checking. Do NOT trigger on MCP server development or general coding tasks.

Install

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

Tool Access

This skill uses the workspace's default tool permissions.

Preview

Analyzes MCP server configuration and recommends improvements based on project stack. Powers `/optimize:mcp` and the MCP configuration dimension of `/optimize:audit`.

Supporting Assets

references/mcp-catalog.md

SKILL.md

Similar Skills

agent-eval

Compares coding agents like Claude Code and Aider on custom YAML-defined codebase tasks using git worktrees, measuring pass rate, cost, time, and consistency.

everything-claude-code

157.7k

agent-harness-construction

Designs and optimizes AI agent action spaces, tool definitions, observation formats, error recovery, and context for higher task completion rates.

everything-claude-code

157.7k

accessibility

Designs, implements, and audits WCAG 2.2 AA accessible UIs for Web (ARIA/HTML5), iOS (SwiftUI traits), and Android (Compose semantics). Audits code for compliance gaps.

everything-claude-code

157.7k

Stats

Parent Repo Stars0

Parent Repo Forks0

Last CommitMar 13, 2026

Actions

View Source View Plugin View on GitHub View README

MCP Advisor

Analyzes MCP server configuration and recommends improvements based on project stack. Powers /optimize:mcp and the MCP configuration dimension of /optimize:audit.

Analysis Workflow

Step 1: Read and Health-Check MCP Configuration

ALWAYS use the bundled script. Do NOT manually read .mcp.json, parse ~/.claude/settings.json, run which, command -v, jq, or any ad-hoc shell commands.

python3 ${CLAUDE_PLUGIN_ROOT}/scripts/mcp_health_check.py --json

The script automatically:

Reads .mcp.json (project) and ~/.claude/settings.json (global)
For each server, extracts: name, command, args, env vars
Checks command existence on system (via which)
Validates required environment variables are set
Estimates tool count from built-in catalog
Calculates token impact (always-loaded vs deferred)

Step 3: Stack-Based Recommendations

Cross-reference detected project stack (from codebase-analyzer) with MCP server catalog.

Reference: ${CLAUDE_PLUGIN_ROOT}/skills/mcp-advisor/references/mcp-catalog.md

Step 4: Context Impact Assessment

The MCP health check script (Step 2) includes token impact estimates. It uses a built-in catalog of known MCP servers to estimate tool counts and calculates:

~150 tokens per always-loaded tool
~20 tokens per deferred tool

Step 5: Security Assessment

Delegate to security-auditor skill's MCP security checks, or inline:

Check enableAllProjectMcpServers
Verify credential handling
Assess trust level

Step 6: Generate Report

## MCP Configuration Report

**Overall Grade**: [A-F] ([score]/100)

### Configured Servers ([count])
| Server | Status | Tools (est.) | Security | Impact |
|--------|--------|-------------|----------|--------|
| [name] | Healthy/Error | XX | OK/Warning | ~XXX tokens |

### Health Issues
| Server | Issue | Fix |
|--------|-------|-----|
| [name] | [description] | [action] |

### Recommended Servers (based on stack)
| Server | Why | Install |
|--------|-----|---------|
| [name] | [detected stack reason] | [install command] |

### Context Impact
- Total always-loaded tools: [count] (~XXX tokens)
- Recommended: Move [count] tools to deferred loading

### Security Summary
[From security-auditor MCP checks]

Scoring Contribution

MCP configuration dimension (10% of total score):

Server health: 30 points
Stack relevance: 25 points
Security posture: 25 points
Context efficiency: 20 points