name: test-rx
description: "Evaluates testing strategy and completeness across 8 dimensions (32 sub-metrics): test pyramid balance, test effectiveness, contract/API testing, UI/visual testing, performance/load testing, test data management, CI integration, and test organization. Produces a scored diagnostic with actionable improvement plans."
triggers:
- "run test-rx"
- "test strategy"
- "evaluate tests"
- "testing audit"
- "test quality review"

Prerequisites

Recommended: lighthouse, pa11y

Check all dependencies: bash scripts/rx-deps.sh or bash scripts/rx-deps.sh --install

test-rx: Testing Strategy Diagnostic

Purpose

Evaluate whether a codebase tests the right things at the right level. This is not about coverage percentages — it is about testing architecture, strategy completeness, and long-term maintainability.

Dimensions (8) and Sub-Metrics (32)

D1: Test Pyramid Balance (15%)

Source: Test Pyramid (Martin Fowler), Practical Test Pyramid (Ham Vocke)

D2: Test Effectiveness (15%)

Source: Mutation Testing (Pitest, Stryker), Google Testing Blog

D3: Contract & API Testing (10%)

Source: Consumer-Driven Contracts (Pact), Schemathesis

D4: UI & Visual Testing (10%)

Source: Chromatic, Percy, Playwright Visual Comparisons

D5: Performance & Load Testing (10%)

Source: k6, Artillery, Lighthouse CI

D6: Test Data Management (15%)

Source: Test Data Management patterns, Factory pattern (fishery, factory_bot)

D7: CI Integration (15%)

Source: Continuous Delivery (Humble & Farley), DORA Metrics

D8: Test Organization & Maintainability (10%)

Source: xUnit Test Patterns (Gerard Meszaros)

Process Overview

1. DISCOVER  ─  Run discover.sh to scan the codebase
2. ANALYZE   ─  4 parallel agents score dimensions
3. SCORE     ─  Aggregate into weighted scorecard
4. PRESCRIBE ─  Generate improvement plan with priorities

Step 1: Discovery

Run the discovery script to collect raw signals from the codebase:

bash ~/.claude/skills/test-rx/scripts/discover.sh "$PROJECT_ROOT"

This produces test-rx-discovery.json with counts, file lists, and pattern matches for all 32 sub-metrics.

Step 2: Parallel Analysis (4 Agents)

Launch 4 agents in parallel, each covering 2 dimensions:

Each agent:
1. Reads the discovery JSON
2. Reads relevant source and test files
3. Scores each sub-metric 0-10 using the grading framework
4. Writes findings to test-rx-d{N}.json

Step 3: Scorecard Aggregation

Combine all dimension scores into the final scorecard:

FINAL SCORE = SUM(dimension_score * dimension_weight)

Step 4: Output

Generate the scorecard in this format:

============================================================
  TEST-RX DIAGNOSTIC SCORECARD
  Project: {project_name}
  Date: {date}
  Final Score: {score}/100 — Grade: {grade}
============================================================

  D1  Test Pyramid Balance     ████████░░  {score}/10  (15%)
  D2  Test Effectiveness       ██████░░░░  {score}/10  (15%)
  D3  Contract & API Testing   ████░░░░░░  {score}/10  (10%)
  D4  UI & Visual Testing      ██░░░░░░░░  {score}/10  (10%)
  D5  Performance & Load       ███░░░░░░░  {score}/10  (10%)
  D6  Test Data Management     ████████░░  {score}/10  (15%)
  D7  CI Integration           ███████░░░  {score}/10  (15%)
  D8  Test Organization        ██████░░░░  {score}/10  (10%)

  WEIGHTED TOTAL: {total}/100

  Grade Scale:
    90-100  A  Exemplary testing strategy
    80-89   B  Strong with minor gaps
    70-79   C  Adequate, clear improvement areas
    60-69   D  Significant strategy gaps
    <60     F  Testing strategy needs overhaul
============================================================

Rules

1. Score only what exists. Do not give credit for intent or plans.
2. Weigh strategy over quantity. 50 good unit tests beat 500 trivial ones.
3. Penalize anti-patterns. Ice cream cone, over-mocking, snapshot-only testing, implementation-coupled tests.
4. Credit test infrastructure. Factories, custom matchers, CI integration are force multipliers.
5. Context matters. A CLI tool does not need visual regression tests. Adjust D4 expectations by project type.
6. Flag risks. No integration tests on a microservice is a critical risk regardless of unit coverage.
7. Be specific. Every finding must reference a concrete file, pattern, or configuration.
8. Improvement plans are mandatory. Every sub-metric scoring below 7 gets a concrete action item.

Auto-Plan Integration

After generating the scorecard and saving the report to docs/audits/:
1. Save a copy of the report to docs/rx-plans/{this-skill-name}/{date}-report.md
2. For each dimension scoring below 97, invoke the rx-plan skill to create or update the improvement plan at docs/rx-plans/{this-skill-name}/{dimension}/v{N}-{date}-plan.md
3. Update docs/rx-plans/{this-skill-name}/summary.md with current scores
4. Update docs/rx-plans/dashboard.md with overall progress

This happens automatically — the user does not need to run /rx-plan separately.