QE Run Workflow

Standalone QE test execution — runs all 6 test layers, classifies failures, and publishes results to Jira with auto-created bug tickets.

12 phasesQE_RUN_COMPLETEQE / SDET

Invocation

bash

# From a Jira ticket
/qe-run PROJ-1234

# Single test layer
/qe-run PROJ-1234 --layer e2e

# From an explicit test plan
/qe-run --from-plan ./TEST_PLAN.md

# Discovery mode (no ticket)
/qe-run "test the search API"

# Re-execute only previous failures
/qe-run --rerun

# Dry run (skip Jira publishing)
/qe-run PROJ-1234 --no-publish

# Commit + PR for new test files
/qe-run PROJ-1234 --pr

Pipeline

Phase Summary

Hover over any phase name to see what it does in detail.

#	Phase	Purpose	Key Output
1	INTAKE	Parse input, fetch ticket, locate test plans	INTAKE.md
2	DISCOVER	Map test landscape, validate plans, identify layers	DISCOVERY.md
3	PLAN	Convert plans into executable commands per layer	EXECUTION_PLAN.md
4	SETUP	Create/update test files, test data, capture baseline	SETUP_RESULT.md
5	EXEC_FUNCTIONAL	Run unit + functional tests (no server)	EXEC_FUNCTIONAL_RESULT.md
6	EXEC_INTEGRATION	Start server, run integration/API tests	EXEC_INTEGRATION_RESULT.md
7	EXEC_E2E	Start server, run E2E + cross-browser tests	EXEC_E2E_RESULT.md
8	EXEC_NONFUNCTIONAL	Performance, security audit, accessibility scan	EXEC_NONFUNCTIONAL_RESULT.md
9	ANALYZE	Aggregate results, classify failures, determine verdict	ANALYSIS.md
10	REPORT	Generate comprehensive QE report	QE_REPORT.md
11	PUBLISH	Post to Jira, create bug tickets, add labels	PUBLISH_RESULT.md
12	LEARN	Persist patterns to shared knowledge stores	—

6 Test Layers

Each layer executes independently. Missing frameworks are recommended, never blocking. Use --layer to run a single layer.

Layer	Scope	Server	Tools	Speed
Functional	Unit + component tests	No	Jest / Vitest / pytest	Fast
Integration	API contracts, service integration	Yes	Supertest / Pact / httpx	Medium
E2E	User journeys, cross-browser	Yes	Playwright / Cypress	Slow
Performance	Load, response time, throughput	Yes	k6 / autocannon / artillery	Medium
Security	Dependency audit, static analysis	No	npm audit / Snyk / ESLint security	Fast
Accessibility	WCAG compliance, ARIA validation	Yes	axe-core / pa11y / Lighthouse	Medium

Failure Classification

The ANALYZE phase classifies every failure by comparing against the baseline captured in SETUP.

REGRESSION

Passed in baseline, fails now. Real issue.

NEW_TEST_FAILURE

Newly created test fails. Bug or test issue.

PRE_EXISTING

Already failing in baseline. Not new.

FLAKY

Inconsistent results across runs.

ENVIRONMENT

Server timeout, missing service. Not code.

Verdicts

Verdict	Meaning	Jira Action
ALL_PASS	Zero failures across all layers	Add qe-verified label
PASS_WITH_KNOWN	Only pre-existing or flaky failures	Add qe-verified label with caveat
REGRESSIONS_FOUND	Tests that passed before now fail	Create Bug tickets, add qe-issues-found label
CRITICAL_FAILURE	Test infrastructure broken	Flag environment issues for investigation
NONFUNCTIONAL_ISSUES	Functional tests pass, NFR findings exist	Create tickets for critical NFR findings

What Gets Posted to Jira

The PUBLISH phase posts a QE report comment, creates Bug tickets for regressions, and adds labels.

## QE Execution Results

**Verdict:** REGRESSIONS_FOUND

**Run date:** 2026-03-06

### Test Summary

- Functional: 48/50 passed (2 regressions)

- Integration: 12/12 passed

- E2E: 8/9 passed (1 flaky)

- Performance: PASS — p95 < 200ms

- Security: WARN — 1 moderate vulnerability

- Accessibility: PASS — 0 critical violations

### Regression Bugs Created

- [PROJ-201] QE Regression: auth.test.ts login returns 401

- [PROJ-202] QE Regression: cart.test.ts total calculation off by 1

**Label added:** qe-issues-found

Story-QA Integration

/qe-run picks up where /story-qa left off. If a TEST_PLAN.md exists from a story-qa run, the PLAN phase converts it into executable commands instead of generating from scratch.

/story-qa→TEST_PLAN.md→/qe-run→QE_REPORT.md

Signal Flow

Signal	Type	Meaning
PHASE_COMPLETE	Advance	Phase done, advance to next in pipeline
SETUP_FAILED	Loopback	Setup failed, retry up to 3 times
BLOCKED_NEEDS_HUMAN	Block	Setup failed 3 times, needs human intervention
EXEC_*_COMPLETE	Advance	Execution layer done (always advances, even with failures)
QE_RUN_COMPLETE	Terminal	Workflow complete, knowledge stores updated