Claude Code Plugin

Autonomous
Test Automation
That Learns From Failure

Built for QA engineers using Claude Code

Triqual writes Playwright tests, runs them, heals failures autonomously, and captures every pattern into a persistent knowledge base. Up to 25 self-healing attempts per test.

/test login— that's it

Start Automating Tests

View on GitHub

Documented Learning Loop — ANALYZE → RESEARCH → PLAN → WRITE → RUN → LEARN

Open Source5 Opus 4.5 AgentsBuilt by a QA EngineerBuilt for Claude Code Developers

See It In Action

One command triggers the entire autonomous pipeline

claude — triqual

What Happens When You Run `/test`

A documented pipeline from command to passing tests — with autonomous healing at every step

GENERATION

One Command, Full Test Suite

Say /test login and Triqual takes over. It explores your app with Playwright MCP, searches Quoth for proven patterns, queries Exolar for failure history, then generates complete spec files with Page Objects.

Supports tickets (/test --ticket ENG-123), descriptions, or feature names. Each run is documented in a structured run log.

Test Generation Flow — from /test command through 6 stages to passing tests

HEALING

Self-Healing Agent Loop

Tests live in .draft/ while the test-healer agent iterates: run, analyze failure, apply fix, run again. Up to 25 attempts with escalating analysis phases. Deep research kicks in at attempt 12.

The failure-classifier categorizes each failure as FLAKE, BUG, ENV, or TEST_ISSUE — no more guessing why tests break.

Draft to Production — .draft/ heals until PASS then promotes to tests/

LEARNING

Every Fix Becomes Team Knowledge

Five specialized Opus 4.5 agents orchestrate the entire workflow. When tests pass, the pattern-learner extracts what worked and proposes it to Quoth — making every future test generation smarter.

Context survives sessions via .triqual/knowledge.md and run logs. Your team benefits from every failure ever fixed.