PlaySwag — API Coverage Analyzer

Analyze an OpenAPI spec against your test suite and generate an HTML report with coverage gaps + ready-made QA automation tasks.

Usage: /playswag [spec-file] [tests-dir] [--js|--ts|--py]

All arguments are optional — missing ones will be asked interactively.

Step 1 — Gather Required Inputs

Spec file

If $ARGUMENTS contains a path to a .json, .yaml, or .yml file — use it as SPEC_FILE.

Otherwise, ask the user:

"What is the path to your OpenAPI/Swagger spec file? (JSON or YAML)"

Verify the file exists:

bash
1test -f "$SPEC_FILE" && echo "found" || echo "not found"

If not found, show the error and ask again.

Tests directory

If $ARGUMENTS contains a second path argument — use it as TESTS_DIR.

Otherwise, auto-detect by checking (relative to spec file's parent dir, then cwd): tests/ → test/ → e2e/ → __tests__/ → specs/ → src/tests/ → . (cwd)

If none of these exist, ask the user:

"Where are your test files? (provide directory path, or press Enter to use current directory)"

Language / Runner

Determine LANG from:

1. --py flag in $ARGUMENTS → LANG=py
2. --ts flag in $ARGUMENTS → LANG=ts
3. --js flag in $ARGUMENTS → LANG=js
4. Auto-detect: check if node is available → LANG=js (recommended default)
5. Auto-detect: check if python3 is available → LANG=py
6. If unclear, ask the user:

"Which runner should I use to analyze?"

• JavaScript (node analyze.js) — runs anywhere with Node.js installed, no extra packages needed ✅
• TypeScript (npx tsx analyze.ts) — requires tsx
• Python (python3 analyze.py) — requires Python 3

Step 2 — Locate Scripts

Find skill scripts dir. Try in order:

bash
1# Project-local (preferred)
2SCRIPTS=".claude/skills/qa-playswag/scripts"
3
4# Global install
5SCRIPTS="$HOME/.claude/skills/qa-playswag/scripts"

Use whichever exists:

bash
1[ -f "$SCRIPTS/analyze.js" ] && echo "found" || echo "not found"

Step 3 — Run the Analyzer

JavaScript (LANG=js) — PRIMARY, recommended

analyze.js is pure CommonJS — runs with any Node.js 14+, zero dependencies needed for JSON specs. For YAML specs it auto-tries js-yaml, yaml, then falls back to python3.

bash
1node "$SCRIPTS/analyze.js" "$SPEC_FILE" "$TESTS_DIR" [options]

TypeScript (LANG=ts) — try runners in order:

bash
1npx --yes tsx "$SCRIPTS/analyze.ts" "$SPEC_FILE" "$TESTS_DIR" [options]

If tsx fails, try ts-node or compile+run. If all fail → auto-fallback to analyze.js.

Python (LANG=py):

bash
1python3 "$SCRIPTS/analyze.py" "$SPEC_FILE" "$TESTS_DIR" [options]

If PyYAML missing and spec is YAML — pip3 install pyyaml first. If Python fails → auto-fallback to analyze.js (if node available).

Multi-spec & URL support

bash
1# Multiple spec files (merge endpoints, dedup by method+path)
2node "$SCRIPTS/analyze.js" spec1.yaml spec2.yaml -- "$TESTS_DIR"
3
4# Spec from URL (fetched and cached in /tmp for 5 min)
5node "$SCRIPTS/analyze.js" https://api.example.com/openapi.json "$TESTS_DIR"

CLI Options (all runtimes)

Exit Codes

CI/CD Example (GitHub Actions)

yaml
1- name: API coverage gate
2  run: node .cursor/skills/qa-playswag/scripts/analyze.js openapi.yaml tests/ --fail-under 80 --json-only

Step 4 — Show Results

After the script runs, read ./playswag-report/summary.json and display:

## PlaySwag Coverage Report

**Spec:** {spec-file}
**Tests:** {tests-dir}

| Metric                    | Value      |
|---------------------------|------------|
| 📊 Endpoint Coverage      | XX% (A/B)  |
| ✗ Uncovered endpoints     | N          |
| 🗑 Deprecated still tested | N          |
| 📋 QA Tasks created       | N          |
| ⚠ Unmatched test calls    | N          |

### Uncovered Endpoints (N) — by priority:

🔴 HIGH (POST/PUT/DELETE/auth):
  - POST /api/users — Create user [users]
  - DELETE /api/orders/{id} — Cancel order [orders]

🟡 MEDIUM (GET):
  - GET /api/reports — List reports [reports]

### Files:
→ HTML Report: ./playswag-report/report.html
→ QA Tasks:    ./playswag-report/tasks.md
→ Badge:       ./playswag-report/playswag-badge.svg
→ Summary:     ./playswag-report/summary.json

Open: open ./playswag-report/report.html

Step 5 — Error Handling

Manual fallback (if all scripts fail):

1. Read spec with Read tool
2. Scan tests with Grep for URL/request patterns
3. Print text summary in chat
4. Write tasks.md manually

Notes

Supported spec formats

Test files detected

.spec.ts/js, .test.ts/js, .e2e.ts/js, .spec.tsx/jsx, test_*.py, *_test.py

HTTP clients detected

Playwright request, axios, fetch, supertest (request(app)), got, ky, httpx, requests, cy.request (Cypress), Node.js http/https, RestAssured (given().when()), aiohttp session, .request('METHOD', path)

Template literal URLs (/api/users/${id}) are partially matched by extracting the static prefix.

Output files

Coverage dimensions

QA task example

From tasks.md:

markdown
1### TASK-001: Cover `POST /api/users`
2
3| Field | Value |
4|-------|-------|
5| **Priority** | High |
6| **Endpoint** | `POST /api/users` |
7| **Auth required** | Yes |
8
9**Acceptance Criteria:**
10- [ ] Happy path -> `201`
11- [ ] Invalid input -> 400/422
12- [ ] Unauthenticated -> 401

Known limitations

• Static analysis only — regex-based; cannot detect dynamic URL construction or runtime-generated paths beyond template literal prefixes.
• Status code coverage is assertion-based (scans for .expect(200) etc.), not runtime-verified.
• Parameter coverage uses name-matching heuristic — a param name appearing in a test file is counted as "used", even if it's unrelated.
• No auth flow verification — authRequired flag comes from spec security definitions, but actual auth testing is not verified.
• YAML requires either js-yaml/yaml npm package, PyYAML, or python3 with PyYAML as fallback.

vs MichalFidor/playswag (npm)

Analyzer behavior

• basePath (Swagger 2.0) and servers[].url path (OAS 3.x) are collected and used when matching test URLs to spec paths (direct match, strip base, or base + template).
• $ref in the spec is resolved (JSON Pointer #/...) before parsing operations, so request body fields in QA tasks are populated from components.schemas / definitions.
• Multi-spec (-- separator): endpoints are merged and deduplicated by method:path.
• URL specs: fetched via curl/wget/node http, cached in /tmp for 5 minutes.
• YAML fallback (when no js-yaml in Node): the temp spec path is passed via PLAYSWAG_YAML_FILE to Python — not embedded in the shell string.
• analyze.js is the safest runner — no build step, no extra packages, works in any Node.js 14+ project.