DEV Community

Furkan ERASLAN
Furkan ERASLAN

Posted on

I Built tensgo: A Zero-Dependency Go CLI & TUI for CI/CD API Load Testing

When adding automated load testing to CI/CD pipelines, I kept running into a trade-off:

  • Lightweight CLI tools like hey and wrk are fast and single-binary, but they lack multi-step stateful flows (e.g., logging in and reusing a Bearer token per virtual user) and strict exit-code latency gates.
  • Heavyweights like k6 or JMeter are powerful, but pulling in a JavaScript runtime or JVM just to fail a build when P95 latency spikes felt excessive.

So I built tensgo (v0.2.0) under Larch Studios—a zero-dependency, single-binary API load and stress testing CLI with an interactive terminal UI written in Go.

What Makes tensgo Different?

  • Strict CI/CD Quality Gates: Fail GitHub Actions or GitLab CI builds automatically with non-zero exit codes if P95 latency or error rates cross your SLA thresholds (--max-p95 250ms --max-error-rate 1.0).
  • OpenAPI / Swagger Auto-Discovery: Point --openapi at your spec URL or local file. tensgo parses endpoints, generates realistic sample payloads, and automatically handles login token extraction.
  • Chaos Profiles & Stateful YAML Flows: Run built-in Spike, Ramp-Up, and Soak chaos profiles, or define multi-step YAML scenarios (Login -> Profile -> Checkout) with per-VU isolated state.
  • Live Bubble Tea TUI + Single-File HTML Export: Monitor live RPS, P50/P95/P99 latencies, and HTTP status distributions in the terminal, and press [h] at any time to export a self-contained HTML report.

Quick Start

Install via Go:

go install github.com/Errtu0/tensgo@latest
Enter fullscreen mode Exit fullscreen mode

Or grab pre-built binaries for Linux, macOS, and Windows from GitHub Releases.

Run a quick load test with a P95 CI gate:

tensgo run https://api.example.com/v1/health -c 50 -d 30s --max-p95 200ms
Enter fullscreen mode Exit fullscreen mode




Links & Interactive Playground

tensgo is MIT-licensed and open to contributions. If you enforce performance budgets in your CI/CD pipelines, I’d love to hear what output formats (JUnit XML, Prometheus metrics, JSON summaries) would make it fit your workflow best!

Top comments (0)