Introduction
Qodex is an AI QA engineer you talk to. In plain-English chat, it can test APIs and web apps, review code changes, probe security, and automate email, performance, and load-testing workflows. Instead of writing every test by hand, describe what you want tested. Qodex explores the target with a real Chrome browser or direct API calls, learns how it behaves, creates runnable Playwright or HTTP scenarios, and saves them for repeatable runs. When a test fails, it triages the evidence as a real bug, stale test, or environment problem.How Qodex works at a high level
Qodex turns a testing brief into reusable test coverage.- Scenarios describe what should be tested, including steps, assertions, tags, and lifecycle state.
- Scripts are generated standard Playwright or HTTP code that can run against your environments.
- Findings are bugs, failures, or vulnerabilities discovered during a run, with severity, reproduction steps, and evidence.
What you can test
AI code review
Review GitHub and GitLab changes, post verified inline findings, test preview deployments, and gate merges with a Check Run.
API testing
Import an OpenAPI spec or Postman collection. Qodex infers auth, creates scenarios, and runs them against any environment.
UI testing
Create intent-driven Playwright scenarios with replay caching, self-healing when pages change, and per-step artifacts on every run.
Security testing
Run continuous OWASP API Top 10 checks. A passing security scenario means the attack was blocked; a failing one includes evidence of the vulnerability.
How tests run
Qodex runs the same saved scenarios in three modes.On-demand
Chat with the agent. Author, run, and triage interactively.
Scheduled
Cron-based recurring runs. Nightly regression, weekly security audit.
Event-driven
CI, deploy hooks, and custom webhooks trigger runs via per-project API keys.
Why replays are cheaper
Most AI QA tools charge per run because every replay is an LLM call. Costs scale with suite size, even when the app has not changed. Qodex uses AI when a test is created or repaired, not every time it runs. The generated script is standard Playwright or HTTP code, parameterized by environment variables, and replayed without calling an AI model. Your nightly regression runs consume execution resources, not AI usage. UI scenarios use natural-language steps resolved against the live page. A successful first run creates an AI-free replay cache; if a page changes enough for the cached path to fail, Qodex re-resolves the step and repairs the test.Where to go next
Quickstart
Connect a project, run your first scan, and read the findings in under five minutes.
How Qodex works
Learn the coordinator model, scan types, and replay architecture.
API testing
Import specs, create scenarios, configure auth profiles, and run API coverage.
PR review
Review pull requests with walkthroughs, inline findings, preview checks, and merge gating.
LLM-readable docs
Every page is also published as.md for LLM ingestion. See llms.txt.