Development

Designs target system boundaries, contracts and tradeoffs from requirements; does not plan tasks or implement.

Documents current architecture from implementation evidence; does not propose a target or audit fitness.

Defines measurable architecture drivers and constraints before system design; edits architecture docs only.

Use when writing any text a human will read: a reply, task summary, explanation, report, plan, spec, or README. Applies to short and technical answers too.

Select refactor candidates with immutable git measurements, evidence-backed module families and temporary collision checks. Use for ranked refactor/test-debt triage, not implementation. DISCOVER saves reports without changing product code; REGISTER requires validated decisions. Worktree names are hints, test LOC is not coverage, and discovery scores are not execution priorities.

Runs a timed mock system design round as the interviewer, then scores it as the coach: seven phases on a 45-minute clock, one follow-up on the weakest area, Mermaid sketches in chat, a six-criterion 0-3 rubric with quoted evidence. Use for interview practice, prep, or grading a practice design.

Inserts stable test ids into production components for each selector gap named by a test plan, following the per-framework insertion policy and never renaming existing ids. Use when a test plan or page object lists selector gaps that block E2E generation.

Turns approved AC/SRS into an executable test plan (scenarios, seed, selector gaps) for the test-loop workflow. Use for independent test-plan generation from stable requirements.

Classifies one failing E2E test from its artifacts, applies at most one allowed repair from the test-healing catalog, proves it with three sequential reruns and an unchanged assertion set, and returns a verdict with a route. Use per failure inside test-loop; never for production code changes.

Draws one evidence-grounded architecture diagram as an editable draw.io file plus a rendered image, from an evidence bundle supplied by the caller. Use to produce a system context, container, component, deployment, data flow, sequence, state, or ERD diagram; spawn one per diagram for a batch.

Captures, masks, compares, and updates screenshot baselines for web and mobile suites, with per-region thresholds and a reviewed-diff rule for every baseline change. Use when a screenshot assertion fails, a baseline needs updating, or visual checks are being added; not for deciding what to verify.

Turn acceptance criteria into an executable test plan (scenarios, seed, selector gaps) before generating E2E code. Use when ACs exist but no runnable test plan does yet.

Failure taxonomy and allowed/forbidden repairs for a failing E2E test. Use when a Playwright/Maestro/Detox/XCUITest/Espresso/Appium test fails and you must decide whether to repair the test or route to a real bug.

Cross-stack selector and test-id policy for web and mobile automation. Use when writing or reviewing E2E/UI test locators, or adding data-testid/accessibility identifiers to components.

Generates Playwright page objects from an executable test plan, one class per screen with ladder-compliant locators and no assertions, wired into a shared fixture. Use when a web scenario needs a page object that does not exist yet, or when locators drift and the page object must be rebuilt.

Quarantines an intermittently failing test behind a ticket with an owner and expiry, assigns a root-cause bucket from isolated reruns, and defines when the test leaves quarantine. Use when a test is a QUARANTINE_CANDIDATE or fails without a code change; not for deterministic failures.

Measures whether an automation suite builds release confidence via feedback-loop length, suite reliability, release cadence, and production escape rate, and emits a release_confidence verdict. Use when judging suite value, ROI, or pre-release trust; not for writing or healing tests.

Modern JavaScript (ES2022+) patterns for clean, maintainable code. Use when working with modern JavaScript features like optional chaining, nullish coalescing, or ESM.

Scores a coding task on Spread, Novelty, and Centrality (SNC 0-6) and maps the tier to autonomy, verification depth, review mode, reviewers, model tier. Use when sizing a fix or feature before planning, choosing fast vs deep review, or deciding whether HARD STOP approval is required.

Runs a timed mock system design round as the interviewer, then scores it as the coach: seven phases on a 45-minute clock, one follow-up on the weakest area, Mermaid sketches in chat, a six-criterion 0-3 rubric with quoted evidence. Use for interview practice, prep, or grading a practice design.

Captures, masks, compares, and updates screenshot baselines for web and mobile suites, with per-region thresholds and a reviewed-diff rule for every baseline change. Use when a screenshot assertion fails, a baseline needs updating, or visual checks are being added; not for deciding what to verify.

Turn acceptance criteria into an executable test plan (scenarios, seed, selector gaps) before generating E2E code. Use when ACs exist but no runnable test plan does yet.

Failure taxonomy and allowed/forbidden repairs for a failing E2E test. Use when a Playwright/Maestro/Detox/XCUITest/Espresso/Appium test fails and you must decide whether to repair the test or route to a real bug.

Cross-stack selector and test-id policy for web and mobile automation. Use when writing or reviewing E2E/UI test locators, or adding data-testid/accessibility identifiers to components.