Language: 日本語English

codspeed Plugin

Author
CodSpeed
Category
Development
Topics
Code Quality, Review & Testing · Monitoring & Observability · Cloud, Deployment & CI/CD
First seen
2026-07-09
Last confirmed
2026-07-13
Explanation last updated
2026-07-12

The explanation below is AI-generated. Please verify it against the sources.

CodSpeed is a continuous performance testing toolkit that detects performance regressions in pull requests and provides tools to diagnose them. According to its homepage, it delivers automated benchmarking, differential flamegraphs, and performance-gated merges so teams can catch slow code before it reaches users. The GitHub README describes a CLI (codspeed) that can benchmark any executable directly or integrate with language-specific harnesses (pytest, criterion, vitest, google-benchmark, and Go's testing package) across Rust, C/C++, Python, Node.js, and Go. It supports three measurement instruments — CPU simulation, memory (eBPF-based), and walltime — and integrates with GitHub Actions, GitLab CI, and Buildkite. The item description states this plugin exposes an MCP server that gives Claude access to benchmarking results, flamegraphs, and comparisons for autonomous performance iteration, though the sources provided do not detail the MCP server's specific commands or capabilities.

Overview

CodSpeed is a hosted performance testing and continuous benchmarking service. Per its homepage and README, it runs benchmarks in CI (or locally), measures execution with low variance using CPU simulation, memory, or walltime instruments, and reports results and regressions as PR comments and status checks.

What you can do with codspeed

  • Benchmark any executable directly with codspeed exec, or integrate with existing test/benchmark frameworks (pytest, criterion, vitest, benchmark.js, google-benchmark, Go's testing package) via codspeed run, per the README
  • Choose among three measurement instruments: CPU simulation (<1% variance), memory profiling (heap allocations via eBPF, for Rust/C/C++), or walltime (real execution time for I/O-heavy or multi-threaded workloads)
  • View differential flamegraphs that pinpoint which lines of code got slower between commits, per the homepage
  • Get automatic PR comments and status checks showing performance impact, and configure merge protection to block PRs that degrade performance beyond a threshold
  • Run benchmarks in CI via GitHub Actions, GitLab CI, or Buildkite using sample workflow configurations provided in the README
  • Define multiple benchmark targets declaratively in a codspeed.yml file for the CLI harness
  • Per the item description, use the CodSpeed MCP server to give Claude granular profiling context (benchmarking results, flamegraphs, comparisons) — specific MCP capabilities beyond this description are unknown

Sources

Original description (English)

CodSpeed is the all-in-one performance testing toolkit. Dive into benchmarking results, flamegraphs, and performance comparisons — give Claude granular profiling context to pinpoint bottlenecks and autonomously iterate on performance via the CodSpeed MCP server.

History of codspeed

Back to list