qa-roles
QA org chart of 15 sharply-scoped role agents across four tiers. Tier 1 IC: exploratory-charter-author, automation-harness-bootstrapper, load-test-plan-designer, security-test-plan-builder, a11y-manual-test-scripter, data-quality-engineer, production-tester. Tier 2 Lead: test-architect, test-effort-estimator, quality-coach, test-quality-coach. Tier 3 Manager: release-engineer, release-cutover-coordinator, qa-manager. Tier 4 Head/Director: head-of-quality. No generic persona-as-scope agents.
Install this plugin
/plugin install qa-roles@testland-qaPart of role bundle: qa-role-leadership
qa-roles
Consolidated QA org chart of 15 sharply-scoped role agents - each ships a specific task for the role, not a job-title persona.
Components
Tier 1 - Hands-on / IC
| Role | Agent | One-line task |
|---|---|---|
| Manual & Exploratory Tester | exploratory-charter-author | Authors SBTM-style session charters (mission, areas, PROOF deliverables, time-box) from a feature spec or risk area. |
| SDET / Automation Engineer | automation-harness-bootstrapper | Scaffolds a full test-automation harness skeleton (folders, fixtures, page-object base, smoke test, CI job) for a repo with no existing suite. |
| Performance Test Engineer | load-test-plan-designer | Designs a tool-agnostic load-test plan mapping SLOs to ramp/soak/spike profiles with pass/fail thresholds. |
| Security Tester / AppSec | security-test-plan-builder | Builds a per-PR security test checklist by mapping the diff's attack surface to OWASP ASVS and Top 10 requirements. |
| Accessibility Specialist | a11y-manual-test-scripter | Produces step-by-step keyboard-navigation and screen-reader manual test scripts mapped to WCAG 2.2 success criteria. |
| Data Quality Engineer | data-quality-engineer | Builds an initial data-quality assertion suite (dbt / GX / Soda) for a single data product from schema and sample. |
| Production Tester | production-tester | Authors and wires a synthetic monitor for one critical user journey, end-to-end including PR creation. |
Tier 2 - Lead / Senior
| Role | Agent | One-line task |
|---|---|---|
| Test Architect | test-architect | Recommends a defensible test pyramid balance and framework choice per repo, backed by current-ratio analysis and ROI math. |
| QA / Test Lead | test-effort-estimator | Estimates testing effort for an epic and proposes a who-tests-what ownership split across the team. |
| Quality Coach (DoD) | quality-coach | Adversarially reviews a story or PR against the team's Definition of Done, tagging each line met / not met / unverifiable. |
| Test Quality Coach | test-quality-coach | Growth-framing coach that scores test files on AAA structure, naming, and conventions for onboarding and ramp-up. |
Tier 3 - Manager / Release
| Role | Agent | One-line task |
|---|---|---|
| Release Engineer | release-engineer | Orchestrates one release runbook (smoke gate, canary deploy, metric thresholds, rollout/rollback) pausing at every human decision point. |
| Release Manager | release-cutover-coordinator | Coordinates a multi-team release cutover - builds the go/no-go checklist, sequences cross-team gates, assigns owners and timeboxes. |
| QA Manager | qa-manager | Generates a weekly RAG quality-status digest from CI history, defect tracker, and flake-quarantine state. |
Tier 4 - Head / Director
| Role | Agent | One-line task |
|---|---|---|
| Head of QA / Director | head-of-quality | Aggregates per-team signals into a portfolio quality roll-up: cross-team KPIs, risk heatmap, and capacity view across squads. |
Install
/plugin marketplace add testland/qa
/plugin install qa-roles@testland-qaAgents
a11y-manual-test-scripter
Produces a manual accessibility test script for a component or page - generates step-by-step keyboard-navigation and screen-reader (NVDA / VoiceOver) test cases mapped to specific WCAG 2.2 success criteria, with expected focus order and announcements. Use when a human needs to manually verify accessibility beyond automated checks; not when statically reviewing code for a11y issues (see accessibility-code-critic in qa-accessibility).
automation-harness-bootstrapper
Scaffolds a test-automation framework skeleton for a repo that has none - given the app's stack and entry points, generates the folder layout, base fixtures, a page-object (or screenplay) base class, one example smoke test, and the CI job that runs it. Use when a team is standing up automated UI/E2E testing from scratch and needs the harness structure before writing tests; not when adding tests to an existing suite (see the *-test-author agents) or auditing an existing framework (see framework-architecture-auditor in qa-test-review).
data-quality-engineer
Builds an initial data-quality coverage suite from scratch for a single data product (one dbt model, one GX-validated table, or one Soda dataset). Reads the schema and a sample, drafts column- and table-level checks, generates the artifacts in whichever engine the project uses, and runs the suite once against the sample. Use proactively when starting coverage on a new model or pipeline. Scope is initial build only: does not audit or gap-review an existing suite.
exploratory-charter-author
Builder agent that authors session-based exploratory testing charters per Bach + Bach SBTM - turns a feature spec / risk area / bug-cluster into a charter card with mission, areas, deliverables (PROOF), and a recommended time-box (60 / 90 / 120 min). Per Bach, exploratory testing is "performing tests while learning things that may influence the testing" - the charter sets the mission while leaving exact steps to the tester's judgment. Use when a feature has too many unknowns to script (new feature / refactor blast-radius / bug cluster) and a session-based exploration is the right approach. Authors the charter only: does not produce post-session debrief templates or coach completed sessions, which belong to the manual-test-debrief skill and the test-quality-coach agent.
head-of-quality
Aggregates per-team quality signals across a multi-squad engineering organisation and produces a portfolio quality review: cross-team KPI roll-up, risk heatmap, capacity and staffing view, quarter-over-quarter trend, and an investment-priority recommendation. Distinct from qa-manager (single-team RAG digest, one repo, one sprint window) - this agent operates at the portfolio layer, reading each team's qa-manager output, OKR set, and release-quality reports rather than raw CI runs. Use when a head of QA, director, or VP needs a portfolio view across multiple teams - e.g., before a quarterly business review, a board update, or a cross-team quality retrospective.
load-test-plan-designer
Designs a load-test plan from a service's SLOs and endpoint inventory - maps each SLO to load scenarios, defines ramp / soak / spike profiles, sets pass/fail threshold expressions, and outputs a tool-agnostic plan ready to implement in k6 or Gatling. Use when planning a performance test before writing the script; not when choosing the load tool (see load-test-tool-selector in qa-load-testing) or bisecting a perf regression (see perf-regression-bisector).
production-tester
Action-taking agent that authors a synthetic monitor for one specific critical user journey, end-to-end - selects the platform per the team's stack (Datadog / Checkly / Pingdom / etc.), generates the script body via accessibility-first locators (per the e2e-selector convention), wires environment-specific config (synthetic test account, test-mode payment processor, multi-region locations, cadence), and outputs both the monitor config + a wire-up PR. Use as a focused per-journey companion to `synthetic-monitor-author` (the broader build-an-X workflow). Does not design the broader monitoring program (platform selection strategy, journey-tier taxonomy, repo layout): for that use synthetic-monitor-author.
qa-manager
Generates a weekly backward-looking quality-status digest for a QA manager - reads CI run history, the defect tracker, and flake-quarantine state, computes pass-rate trend, escape-defect rate, and flake debt, and emits a one-page red / amber / green status per area. Use weekly before a quality review, or when a manager asks where quality stands this sprint. Composes existing signals into a status doc; does not itself run tests or triage defects, and does not set targets, OKRs, or thresholds. For defining forward-looking quarterly quality goals use head-of-quality.
quality-coach
Adversarial reviewer that critiques a story / PR / Increment against the team's Definition of Done - reads the team's `docs/definition-of-done.md` (or whichever path the project uses), then walks each DoD line and tags it `met` / `not met` / `unverifiable`, calling out which lines lack evidence in the PR. Per the Scrum Guide DoD is "a formal description of the state of the Increment when it meets the quality measures required for the product"; this agent makes adherence visible. Use during PR review or before a story moves to "done" - surfaces the unmet items so the team doesn't ship work that doesn't actually meet quality.
release-cutover-coordinator
Coordinates the org-level cutover during a live release window across multiple teams - sequences multi-team dependency order, assigns gate owners and timeboxes, builds the go/no-go checklist, and documents explicit rollback decision points per gate. Use when multiple teams are cutting over in the same window and need a single cross-team sequencing runbook; not for the pre-window readiness gate (see release-readiness-checker in qa-process) or single-service runbook execution (see release-engineer).
release-engineer
Builder/scaffolder agent that orchestrates the release runbook for one specific release - reads the team''''s `docs/release-runbook.md` (or planned `qa-process` Plugin 16''''s runbook template), executes its steps in order (smoke suite gate → canary deploy with metric thresholds → rollout / rollback decision), records pass/fail per step, and emits a release report. Implements the canary release pattern per [canary-release][cr]: "rolling out the change to a small subset of users before rolling it out to the entire infrastructure". Use as the conductor for a release; never auto-merges or auto-rolls-out - always pauses at decision points for human approval.
security-test-plan-builder
Builds a per-PR security test checklist from a change's attack surface - reads the diff, maps touched surfaces (authentication, input handling, file upload, deserialization, access control) to the relevant OWASP ASVS verification requirements and Top 10 categories, and emits a targeted manual + automated security test list. Use when scoping security tests for a specific change before findings exist; not when triaging existing SAST/DAST findings (see sast-finding-triager, dast-finding-triager).
test-architect
Action-taking agent that, given a single repo + a recent change set, recommends a defensible test pyramid balance (unit / integration / E2E split) and a testing-framework choice - reads the existing test-suite to compute current ratios per [test-pyramid][tp] thinking, examines the change set to see whether it's the right shape (UI-heavy / service-heavy / data-heavy), and emits a written rationale for the recommendation including ROI math (cost vs failure-detection lift). Use as a per-repo pre-investment review before the team commits to a new framework or shifts the pyramid balance.
test-effort-estimator
Estimates testing effort for an epic and proposes an ownership split - given the epic's stories and change shape, classifies test work by layer and risk, produces a per-area effort estimate with stated assumptions, and recommends who-tests-what across the team. Use when planning test capacity for upcoming work; not when selecting which tests to run for a given change (see risk-based-test-selector) or planning risk coverage (see risk-based-test-planner in qa-process).
test-quality-coach
Growth-framing coach for **test-design quality** - scores each test file in the diff on AAA structure, naming, single-responsibility, magic numbers, and slow setup to improve how tests are designed, not to enforce a Definition of Done. Differs from `quality-coach` (DoD-adherence enforcer) - this agent never blocks a PR; it coaches test-design thinking (coverage heuristics, convention application, growth path) for onboarding and ramp-up. Differs from `test-code-critic` (same conventions, adversarial pass/fail framing) - this agent uses **growth framing** ("here's what to improve next time").