I Built a Testing Tool That Analyzes Code Across 21 Dimensions — Here is Why

The Problem

I was shipping AI-generated code fast. Really fast. But here is the thing — nobody was testing it properly.

Sure, I had ESLint. I had Jest for unit tests. But when I looked at the testing landscape, I realized something:

• Snyk checks security vulnerabilities (1 dimension)
• SonarQube checks code quality (2-3 dimensions)
• TestRail manages test cases (manual, 1 dimension)

Nobody checks if your API can handle thousands of AI agents hitting it simultaneously. Nobody checks your DORA metrics. Nobody does supply chain audits, N+1 query detection, or agentic scale prediction.

So I built TestForge — a testing platform that analyzes code across 21 dimensions.

What Makes It Different

Standard Dimensions

• Security scanning (SAST)
• Unit test coverage
• Load & performance
• Accessibility (WCAG)

Strategic Dimensions (Only TestForge)

🤖 Agentic Scale Prediction — World first. Simulates thousands of AI agents hitting your API. Predicts failure patterns, bottlenecks, and cascading failures.

📊 DORA Metrics — Deployment frequency, lead time, MTTR, change failure rate.

🔗 Supply Chain Audit — Cross-references deps against CVE databases with fix versions.

🔍 N+1 Query Detection — Finds DB queries inside loops that ORMs hide from you.

👁️ Vision & Goal Alignment — Observability, feature flags, analytics, CI/CD maturity.

📦 Dead Code + License — Unused deps, dead exports, GPL/copyleft risks.

How It Works

1. Go to testforge.run
2. Paste any public GitHub repo URL
3. Get a full 21-dimension report in ~30 seconds

No config. No setup. No YAML files.

Results from testing tinyhttp/malibu

Security           4 findings  ████░░░░░░░░░░░░░░░░
Unit Tests         67%         █████████████░░░░░░░
Agentic Scale      24/100      ████░░░░░░░░░░░░░░░░  ⚠️
DORA               30/100      ██████░░░░░░░░░░░░░░  ⚠️
OWASP Coverage     20%         ████░░░░░░░░░░░░░░░░  ⚠️

• 3 auth middleware gaps (HIGH)
• No rate limiting — critical for agentic workloads
• No observability or product analytics
• 34 undocumented API endpoints

Try It

# Web platform
open https://testforge.run

# MCP IDE integration
npx @whitenoisenpm/testforge-mcp install

# Self-hosted (5 min)
git clone https://github.com/t4tarzan/testforge
cd testforge/mcp-server && flyctl launch --now

Try it: testforge.run
GitHub: t4tarzan/testforge
npm: npx @whitenoisenpm/testforge-mcp install

Built by t4tarzan