Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Set up MCP Inspector for debugging and testing MCP servers with request logging, response inspection, and protocol validation.
.claude/skills/a5c-ai-mcp-inspector-integration/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 24% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 66% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -43% | 0% |
Set up MCP Inspector for debugging and testing MCP servers.
Invoke this skill when you need to:
| Parameter | Type | Required | Description | |-----------|------|----------|-------------| | serverPath | string | Yes | Path to MCP server entry | | transport | string | No | Transport type (stdio, sse) | | logging | boolean | No | Enable verbose logging |
json{ "mcpServers": { "my-server": { "command": "node", "args": ["dist/index.js"], "env": { "DEBUG": "mcp:*", "NODE_ENV": "development" } } } }
typescript#!/usr/bin/env node import { createServer } from './server'; // Enable debug logging process.env.DEBUG = 'mcp:*'; // Log all stdin/stdout for debugging const originalWrite = process.stdout.write.bind(process.stdout); process.stdout.write = (chunk: any, ...args: any[]) => { if (process.env.MCP_DEBUG_LOG) { const fs = require('fs'); fs.appendFileSync( process.env.MCP_DEBUG_LOG, `[OUT] ${new Date().toISOString()} ${chunk}\n` ); } return originalWrite(chunk, ...args); }; process.stdin.on('data', (chunk) => { if (process.env.MCP_DEBUG_LOG) { const fs = require('fs'); fs.appendFileSync( process.env.MCP_DEBUG_LOG, `[IN] ${new Date().toISOString()} ${chunk}\n` ); } }); // Start server createServer();
json{ "scripts": { "dev": "tsx watch src/index.ts", "debug": "MCP_DEBUG_LOG=./debug.log tsx src/index.ts", "inspect": "npx @anthropic/mcp-inspector node dist/index.js", "test:mcp": "npx @anthropic/mcp-inspector --test node dist/index.js" } }
typescript// tests/mcp-scenarios.ts export const testScenarios = [ { name: 'List Tools', request: { jsonrpc: '2.0', method: 'tools/list', id: 1, }, validate: (response: any) => { return response.result?.tools?.length > 0; }, }, { name: 'Call Tool - search_files', request: { jsonrpc: '2.0', method: 'tools/call', params: { name: 'search_files', arguments: { pattern: '*.ts', path: './src' }, }, id: 2, }, validate: (response: any) => { return !response.error && response.result?.content; }, }, { name: 'List Resources', request: { jsonrpc: '2.0', method: 'resources/list', id: 3, }, validate: (response: any) => { return Array.isArray(response.result?.resources); }, }, ]; // Run test scenarios export async function runScenarios( send: (msg: any) => Promise<any> ): Promise<void> { for (const scenario of testScenarios) { console.log(`Testing: ${scenario.name}`); const response = await send(scenario.request); const passed = scenario.validate(response); console.log(` ${passed ? '✓ PASS' : '✗ FAIL'}`); if (!passed) { console.log(' Response:', JSON.stringify(response, null, 2)); } } }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 14,078 | 10,986 | -22% | 1 | 1 | 0% | 2,898 | 3,594 | +24% | 0 | 0 | — |
case-02 | fail→pass | 21,808 | 8,607 | -61% | 1 | 1 | 0% | 2,167 | 3,188 | +47% | 0 | 0 | — |
case-03 | fail→pass | 25,607 | 11,842 | -54% | 1 | 1 | 0% | 2,268 | 3,776 | +66% | 0 | 0 | — |
case-04 | fail→fail | 6,758 | 4,161 | -38% | 1 | 1 | 0% | 1,325 | 1,892 | +43% | 0 | 0 | — |
case-05 | fail→pass | 6,763 | 2,791 | -59% | 1 | 1 | 0% | 1,230 | 1,553 | +26% | 0 | 0 | — |
case-06 | fail→fail | 16,215 | 12,473 | -23% | 1 | 1 | 0% | 2,926 | 3,526 | +21% | 0 | 0 | — |
case-07 | pass→pass | 14,031 | 5,761 | -59% | 1 | 1 | 0% | 2,605 | 2,345 | -10% | 0 | 0 | — |
case-08 | fail→fail | 6,484 | 5,548 | -14% | 1 | 1 | 0% | 1,328 | 2,285 | +72% | 0 | 0 | — |
case-09 | fail→fail | 6,129 | 2,661 | -57% | 1 | 1 | 0% | 1,264 | 1,704 | +35% | 0 | 0 | — |
case-10 | fail→fail | 9,383 | 4,315 | -54% | 1 | 1 | 0% | 1,845 | 2,043 | +11% | 0 | 0 | — |
case-11 | fail→fail | 12,193 | 9,195 | -25% | 1 | 1 | 0% | 2,562 | 3,215 | +25% | 0 | 0 | — |
case-12 | fail→fail | 9,021 | 6,533 | -28% | 1 | 1 | 0% | 1,894 | 2,518 | +33% | 0 | 0 | — |
case-13 | fail→fail | 2,914 | 3,018 | +4% | 1 | 1 | 0% | 547 | 1,710 | +213% | 0 | 0 | — |
case-14 | fail→pass | 15,801 | 2,789 | -82% | 1 | 1 | 0% | 2,941 | 1,676 | -43% | 0 | 0 | — |
case-15 | fail→fail | 5,365 | 4,544 | -15% | 1 | 1 | 0% | 908 | 1,936 | +113% | 0 | 0 | — |
case-16 | fail→fail | 7,768 | 4,851 | -38% | 1 | 1 | 0% | 1,567 | 1,933 | +23% | 0 | 0 | — |
case-17 | fail→pass | 12,488 | 2,558 | -80% | 1 | 1 | 0% | 2,094 | 1,574 | -25% | 0 | 0 | — |
case-18 | fail→pass | 11,069 | 2,996 | -73% | 1 | 1 | 0% | 1,975 | 1,732 | -12% | 0 | 0 | — |
case-19 | fail→fail | 13,425 | 10,062 | -25% | 1 | 1 | 0% | 2,959 | 3,315 | +12% | 0 | 0 | — |
case-20 | fail→pass | 12,921 | 10,425 | -19% | 1 | 1 | 0% | 2,918 | 3,427 | +17% | 0 | 0 | — |
case-21 | fail→pass | 12,494 | 9,974 | -20% | 1 | 1 | 0% | 2,693 | 3,455 | +28% | 0 | 0 | — |
case-22 | fail→fail | 5,228 | 7,221 | +38% | 1 | 1 | 0% | 1,205 | 2,806 | +133% | 0 | 0 | — |
case-23 | fail→fail | 13,554 | 13,180 | -3% | 1 | 1 | 0% | 2,736 | 4,233 | +55% | 0 | 0 | — |
case-24 | fail→fail | 4,555 | 4,070 | -11% | 1 | 1 | 0% | 860 | 1,794 | +109% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted. The headline lift of +38 percentage points is the difference between those two pass rates over the 24 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.