skills.rest
sourcegraph
sourcegraphOFFICIAL
Explore all skills by this author →
Website𝕏@sourcegraphGitHub

mcp-audit

Analyze MCP tool usage patterns and performance deltas from task_metrics.json files.

32|4|Updated Feb 1, 2026
Search Tags:#performance analysis#tool usage#mcp audit#code agent#benchmark analysis#sourcegraph
One-click install
npx skills add https://github.com/sourcegraph/CodeScaleBench --skill mcp-audit-sourcegraph
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mcp-audit
Source: https://github.com/sourcegraph/CodeScaleBench/tree/main/skills/mcp-audit
Command: npx skills add https://github.com/sourcegraph/CodeScaleBench --skill mcp-audit-sourcegraph

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill helps understand how and where the MCP (Sourcegraph) tool is used within benchmark runs, identifying its benefits, drawbacks, and areas of underutilization.

Core Features & Use Cases

  • Usage Analysis: Classifies tasks based on MCP adoption (zero-MCP vs. used-MCP).
  • Performance Deltas: Computes reward and time differences conditioned on MCP usage.
  • Root Cause Investigation: Helps investigate why MCP might not be used or why it leads to regressions.
  • Use Case: A researcher wants to know if using Sourcegraph's MCP tools actually improves coding agent performance on complex tasks, or if it introduces unnecessary overhead.

Quick Start

Run the mcp audit script to analyze tool usage patterns and performance deltas.

Frequently Asked Questions about mcp-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I analyze MCP tool usage impact on code agent benchmark performance?▼

To analyze benchmark performance impact, this Skill processes task_metrics.json files to categorize tasks by MCP adoption and calculates reward and time deltas across benchmark runs.

What is the best way to identify negative performance regressions caused by MCP adoption?▼

Identifying negative performance regressions from MCP adoption involves calculating performance deltas between zero-MCP and used-MCP tasks to pinpoint where Sourcegraph tools introduce overhead or fail to improve agent outcomes.

How do I benchmark Sourcegraph MCP tool usage against tasks where it was not used?▼

You can benchmark Sourcegraph MCP tool usage by running the mcp_audit.py script, which classifies tasks into zero-MCP and used-MCP categories and computes the reward and time differences between them.

Why does my coding agent underutilize MCP tools during benchmark runs?▼

Your coding agent may underutilize MCP tools due to task constraints or configuration issues. This Skill helps investigate root causes by analyzing usage patterns and identifying where MCP was not adopted.

Do I need task_metrics.json files to run an MCP audit analysis?▼

Yes, you need task_metrics.json files because the mcp_audit.py script requires them as the primary data source for collecting metrics, categorizing MCP adoption, and calculating performance deltas.

Related Skills

Explore More Software Engineering Skills→
Hex-4Hex-4

session-report

Analyze Claude Code session transcripts to generate HTML usage reports.

Community
Advanced
him55710-sudohim55710-sudo

benchmark

Measure page performance metrics and compare them against baselines to detect regressions.

Community
Advanced
jasonmeansjasonmeans
2

sourcegraph-search

Integrates with Sourcegraph to search code, commits, diffs, symbols, and DeepSearch across repositories.

Community
Advanced
sourcegraphsourcegraph
32

Evaluate Benchmark Traces

Evaluate AI agent benchmark traces for data integrity, output quality, and efficiency.

Official
Advanced