원클릭으로 Manus에서 모든 스킬 실행

orbit-gauntlet

Run the full Orbit gauntlet — 11 sequential checks on a WordPress plugin (PHP lint, PHPCS, PHPStan, asset weight, i18n, Playwright, Lighthouse, DB profile, competitor compare, UI perf, AI skill audits, PM UX). Three modes — `quick` (3-5 min, dev loop), `full` (30-45 min, RC pass), `release` (45-60 min, WP.org submission). Use when the user says "run gauntlet", "audit my plugin", "full QA", "before release", or any unscoped "check my plugin" request.

Manus에서 실행

스타1

포크0

업데이트2026년 6월 2일 10:12

출처

adityaarsharma

adityaarsharma/orbit

GitHub 저장소 열기 Creator 저장소 보기

설치 명령

다운로드

Manus에서 실행

SKILL.md

readonly

이 저장소의 다른 Skills

같은 저장소

orbit-compat-polylang

adityaarsharma/orbit

Polylang compatibility audit — pll_register_string, pll_current_language, custom-post-type translation, language switcher hooks, REST + WP-CLI integration. Polylang has free + Pro versions; covers both. Use when the user says "Polylang compat", "Polylang translate", "WPML alternative", or coexisting with Polylang.

2026-06-021

orbit-compat-wpml

adityaarsharma/orbit

WPML compatibility audit — translatable strings via icl_t / wpml_register_string, custom-post-type translation, taxonomy translation, language switcher hooks, current-language detection, sitemap-per-language, and the wpml-config.xml registration file. Use when the user says "WPML compat", "WPML translate strings", "translation plugin", or before customer asks "does this work with WPML?".

2026-06-021

orbit-i18n-js-parity

adityaarsharma/orbit

PHP↔JS label parity audit — verifies every label key consumed by JS via `wp_localize_script` / `wp_set_script_translations` is actually emitted from PHP. Catches the silent-English-fallback bug where JS reads `myObject.someLabel` but PHP never put `someLabel` into the localized object. Use when the user says "JS label", "wp_localize_script", "Elementor editor label", "Gutenberg block label", "missing translation in JS", or before any release that touches block editor / Elementor / admin JS.

2026-06-021

orbit-i18n-runtime

adityaarsharma/orbit

Runtime i18n correctness audit — catches the non-Latin data corruption bugs that gettext-style audits miss. Scans every wp_json_encode / json_encode call in paths that store post_meta, options, REST output, or LLM/external-API request bodies; flags any missing JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES flag. Also verifies Content-Type charset=utf-8 on REST responses. Use when the user says "i18n runtime", "Unicode encoding", "non-Latin characters", "Turkish broken", "CJK garbage", "data corruption on save", or before any release that ships to non-English locales.

2026-06-021

orbit-i18n-translator-currency

adityaarsharma/orbit

Translator .po file currency audit — for every shipped /languages/<plugin>-<locale>.po file, runs `msgmerge` against the current .pot and reports the percentage of msgids that are missing, fuzzy, or obsolete vs the current source code. Catches the silent-English-fallback bug where a translator delivered 100% translation against an OLD pot, but new strings shipped since. Use when the user says "translator currency", "stale po", "translation drift", "translator delivered translation", "ship translations", or before any release that has shipped .po/.mo files.

2026-06-021

orbit-code-quality

adityaarsharma/orbit

Code-quality reviewer for WordPress plugins — finds dead code, complexity hotspots, error-handling gaps, type-safety issues, AI-hallucination risks (made-up WP function names, wrong sanitize choice, missing nonce, class-name case drift, method API drift), AND the WordPress runtime traps that only break on a live install (Settings API cross-nulling, DISABLE_WP_CRON assumptions, conditional add_rewrite_rule, bulk option restore wipes, %currentyear% literals in third-party SEO plugin meta, Gutenberg block-comment false-positive text checks, tab/router slug mismatch, Pro/Free shadow-class conflicts, cross-plugin filter timing). Use when the user says "code quality", "vibe code review", "review AI-generated code", "find dead code", "complexity audit", "why didn't Orbit find this", or after merging a Cursor/Copilot-assisted PR.

2026-05-291

name	orbit-gauntlet
description	Run the full Orbit gauntlet — 11 sequential checks on a WordPress plugin (PHP lint, PHPCS, PHPStan, asset weight, i18n, Playwright, Lighthouse, DB profile, competitor compare, UI perf, AI skill audits, PM UX). Three modes — `quick` (3-5 min, dev loop), `full` (30-45 min, RC pass), `release` (45-60 min, WP.org submission). Use when the user says "run gauntlet", "audit my plugin", "full QA", "before release", or any unscoped "check my plugin" request.
argument-hint	--plugin <path> --mode quick\|full\|release

🪐 orbit-gauntlet — The full pipeline

The flagship Orbit command. One invocation, every quality angle.

Smart-agentic mandate. --mode full and --mode release run every step end-to-end, no cherry-picking. --skip N is debug-only — using it for any release-track run requires a brain note (orbit/08-release) recording the reason. Smart-agentic = aggressive; conservative shipped 5 RankReady i18n bugs. See agents/_SMART-AGENTIC-MANDATE.md.

Quick start

# Daily dev loop — fast feedback (~3-5 min)
bash scripts/gauntlet.sh --plugin ~/plugins/my-plugin --mode quick

# Pre-PR / beta release — everything (~30-45 min)
bash scripts/gauntlet.sh --plugin ~/plugins/my-plugin --mode full

# Day-of-release — full + WP.org plugin-check (~45-60 min)
bash scripts/gauntlet.sh --plugin ~/plugins/my-plugin --mode release

Exit code: 0 = release ready · 1 = blockers found.

Mode comparison

Step	quick	full	release
1. PHP lint	✓	✓	✓
1a. Release metadata	✓	✓	✓
1b. Zip hygiene	—	✓	✓
2. PHPCS (WPCS + VIP)	✓	✓	✓
3. PHPStan level 5	✓	✓	✓
4. Asset weight	✓	✓	✓
5. i18n + POT (gettext)	—	✓	✓
5a. i18n runtime (UNESCAPED_UNICODE)	—	✓	✓
5b. i18n JS parity (wp_localize_script)	—	✓	✓
5c. i18n translator currency (.po drift)	—	✓	✓
5d. Polylang + WPML endpoint awareness	—	✓	✓
6. Playwright (smoke)	✓	full suite	full suite
7. Lighthouse	—	✓	✓
8. DB profiling	—	✓	✓
9. Competitor compare	—	✓	✓
10. UI / editor perf	—	✓	✓
11. 6 AI skill audits	—	✓	✓
12. PM UX (spell + label)	—	✓	✓
WP.org plugin-check	—	—	✓
Live CVE feed	—	—	✓
Ownership-transfer detect	—	—	✓

Reading the output

After every run:

reports/
├── qa-report-<timestamp>.md         ← markdown summary (start here)
├── playwright-html/index.html       ← interactive E2E report
├── skill-audits/index.html          ← tabbed AI audit (full + release only)
├── lighthouse/lh-<timestamp>.json   ← performance data
└── pm-ux/pm-ux-report-<timestamp>.html  ← PM-friendly view

Open the master index:

python3 scripts/generate-reports-index.py
open reports/index.html

Severity triage

Level	Action
Critical	Block release. Fix today.
High	Block release. Fix in this PR.
Medium	Fix if <30 min. Otherwise log.
Low / Info	Log. Defer.

Any Critical or High in modes full / release → exit code 1.

Common failures

`wp-env not running on port 8881`

Site isn't up. Run /orbit-docker-site first.

`phpcs: command not found`

Power tools not installed. Run /orbit-install.

`Playwright tests failed — auth.setup.js`

Admin cookies expired. Re-run:

WP_TEST_URL=http://localhost:8881 npx playwright test \
  tests/playwright/auth.setup.js --project=setup

`qa.config.json not found`

First-time setup needed. Run /orbit-setup.

Step hangs > 10 min

Probably the AI skill audits in --mode full waiting on Claude Code. Check reports/skill-audits/*.md to see if any wrote partial output. Kill + re-run.

What each step actually does

Detailed in docs/04-gauntlet.md. Quick reference:

Step 1   PHP Lint           → syntax errors in every .php file (~10s)
Step 1a  Release Metadata   → header, readme.txt, version parity, license, HPOS, WP compat (~20s)
Step 1b  Zip Hygiene        → dev files, forbidden functions, supply-chain audit (~30s)
Step 2   PHPCS              → WordPress + VIP coding standards (~30s)
Step 3   PHPStan            → static analysis (level 5) (~45s)
Step 4   Asset Weight       → JS/CSS bundle sizes (~5s)
Step 5   i18n / POT         → translatable strings + text domain check (~20s)
Step 5a  i18n runtime       → JSON_UNESCAPED_UNICODE in storage + REST charset (~10s)
Step 5b  i18n JS parity     → PHP wp_localize_script keys vs JS reads (~10s)
Step 5c  i18n translator    → per-locale .po staleness vs current POT (~15s)
Step 5d  Multilingual compat→ Polylang + WPML language-aware custom endpoints (~20s)
Step 6   Playwright Tests   → functional + visual regression + flow videos (~3 min)
Step 7   Lighthouse         → Core Web Vitals scores (~1 min)
Step 8   DB Profiling       → query count + slow query log + memory + cron + GDPR (~2 min)
Step 9   Competitor         → side-by-side comparison of competitor plugins (~3 min)
Step 10  UI Performance     → editor load time (Elementor/Gutenberg) + frontend TTFB (~2 min)
Step 11  Claude Skills      → 6 parallel AI audits (security, perf, DB, a11y, standards, quality)
Step 12  PM UX Audit        → spell-check + guided experience score + label benchmarking

After failure — drill into a specific layer

Layer that failed	Skill
PHPCS / WPCS	`/orbit-wp-standards`
Security / SAST	`/orbit-wp-security`
Performance	`/orbit-wp-performance`
Database	`/orbit-wp-database`
Accessibility	`/orbit-accessibility`
i18n (gettext)	`/orbit-i18n`
i18n runtime (Unicode encoding)	`/orbit-i18n-runtime`
i18n JS parity (silent English fallback)	`/orbit-i18n-js-parity`
i18n translator currency (.po drift)	`/orbit-i18n-translator-currency`
Polylang compat (incl. custom endpoints)	`/orbit-compat-polylang`
WPML compat (incl. wpml-config.xml currency)	`/orbit-compat-wpml`
Playwright	`/orbit-playwright`
Lighthouse	`/orbit-lighthouse`
Editor perf	`/orbit-editor-perf`
PM UX (spell + label)	`/orbit-pm-ux-audit`
WP.org plugin-check	`/orbit-plugin-check`

Each sub-skill goes deeper than the gauntlet step.

Variations

# Override port
WP_TEST_URL=http://localhost:8888 bash scripts/gauntlet.sh --plugin . --mode full

# Skip a specific step (debug-only — never for release)
bash scripts/gauntlet.sh --plugin . --mode full --skip 9

# Run only steps 1-5 (fast PHP-only audit)
bash scripts/gauntlet.sh --plugin . --mode quick --only 1-5

# Output JSON instead of markdown
bash scripts/gauntlet.sh --plugin . --mode full --format json

CI / GitHub Actions

gauntlet.sh is CI-ready out of the box — exit 0/1, deterministic output:

- name: Orbit gauntlet
  run: bash scripts/gauntlet.sh --plugin . --mode full
- name: Upload reports
  if: always()
  uses: actions/upload-artifact@v4
  with: { name: orbit-reports, path: reports/ }

Full template: docs/15-ci-cd.md.