Description
Testers.ai makes CARBON, an AI testing harness designed to sit inside the coding agents developers already use rather than replace them. The pitch is that CARBON turns an AI coding agent into one with a QA team attached: it adds testing knowledge, repeatable QA workflows, and specialist AI sub-agents to tools like Claude, Codex, Cursor, and Antigravity, connected through the Model Context Protocol (MCP).
Read More
CARBON follows a risk-based testing loop: it analyzes code and existing coverage to flag risky areas, plans tests around real user journeys and coverage gaps, generates and runs test cases, and produces a confidence score backed by evidence rather than a pass or fail count alone. Specialist AI sub-agents focus on specific concerns, including security, OWASP checks, accessibility and WCAG compliance, usability, performance, localization, data integrity, forms, and checkout flows, among others. The tool is invoked from inside the coding agent with slash commands, for example /carbon to run the full loop, or /carbon-security for a focused security pass, and there are more than 50 commands covering everything from AI evaluation design to release-risk review.
CARBON Local Harness runs inside an existing coding agent and reads the project's own code and tests; it is free to use and requires only the developer's existing agent subscription or API access. CARBON Cloud works differently: given a URL, it autonomously explores a live site, exercises real user journeys, and returns bugs with supporting evidence, without needing the project's source code. A free sample cloud check is available without an account or credit card, subject to service limits.
CARBON is built by Jason Arbon, whose background includes QA and testing work at Google, Microsoft, Bing, and Applause, and who has published books on software testing. The product is aimed at development teams that want structured, evidence-based QA layered onto AI-assisted coding rather than a separate manual testing process, and at teams releasing AI features who need to evaluate models, prompts, and agent behavior alongside traditional functional testing.
- CategoryAI Code Review & Testing
- Pricing ModelFreemium
- DeploymentCloud
- PlatformsLocal, Cloud
Testers.ai's Core Features
CARBON Local Harness runs inside Claude, Codex, Cursor, and Antigravity via MCP
CARBON Cloud tests a live URL autonomously with no source code access needed
Risk-based testing loop: analyze code, plan coverage, generate tests, score confidence
24 specialist AI sub-agents for areas like security, accessibility, and performance
50-plus slash commands for focused testing tasks
AI evaluation tools for models, prompts, RAG, and agent behavior
Release-readiness recommendations backed by evidence
Free local harness; free sample check for the cloud product
How to use Testers.ai?
Install CARBON: Add the CARBON plugin to Claude, Codex, Cursor, or Antigravity via MCP
Run the full loop: Use /carbon to analyze code, plan tests, and gather evidence
Focus on an area: Run a specific command like /carbon-security or /carbon-accessibility
Review the confidence score: Check what CARBON flags as ready or not ready
Test a live site: Give CARBON Cloud a URL to explore for release risks without local setup
Iterate: Re-run /carbon-more or /carbon-auto to expand coverage as code changes
Testers.ai's Use Cases
- AI-assisted QA in coding agents
- Pre-release risk checks
- Live site testing
- AI feature evaluation
