ReadyVibe-Skills

ReadyVibe-Skills

Production and launch skills for vibe-coded websites.

Your AI coding agent built a site that looks finished. These skills teach that agent to find what still stands between the site and a responsible launch: it inspects what the project actually does, fixes what is safe to fix, verifies each change, and tells you plainly what needs a human or a lawyer.

Quick start

npx skills add moh-obaida/ReadyVibe-Skills 

Then, in your coding agent, inside your project:

Make this ready to launch.

That is all. the agent uses the skills and inspects your project, decides which launch areas apply, checks only those, fixes safe issues in your existing design, verifies the fixes, and ends with a short report: READY · FIXED · BLOCKERS · REVIEW REQUIRED · UNVERIFIED · NEXT ACTION.

No ReadyVibe CLI, package, account, or service is required. A few browser-based checks use Playwright if your project has it; a skill tells you when it cannot run one.

Install options

npx skills add moh-obaida/ReadyVibe-Skills --list                      # see every skill
npx skills add moh-obaida/ReadyVibe-Skills --skill launch-all          # the broad entry point
npx skills add moh-obaida/ReadyVibe-Skills --skill compliance-all      # privacy, consent, email, age, rights, legal pages
npx skills add moh-obaida/ReadyVibe-Skills --skill admin-dashboard     # build a full admin tailored to your app
npx skills add moh-obaida/ReadyVibe-Skills --skill faq-readiness       # an FAQ from real questions and verified answers
npx skills add moh-obaida/ReadyVibe-Skills                             # every skill there is

Add --agent <name> (for example --agent claude-code) to install for one agent only. These commands were verified against Skills CLI 1.7.0.

Pin to a release

Skills are instructions your agent runs, so pin an install to a release if you do not want it to change underneath you:

npx skills add moh-obaida/ReadyVibe-Skills#v1.0.1 

The pinned form (#v1.0.0, a tag, or a full commit SHA) was verified end to end against the v1.0.1 release: listing, isolated installs, the default install layout, and --all. See the releases and the CHANGELOG.

Standalone skills, and when specialists go deeper

Every skill works on its own after install. Its helper scripts and shared references travel inside its own folder.

Where a skill relies on another skill's method (launch-all may need the SEO lane, privacy-policy may need the data-flow lane), it declares that skill as a companion:

• Companion installed: the agent uses its full method. That is the deeper result.
• Companion not installed: the skill carries a short reduced-depth fallback for that lane, and its report says which lanes ran inline at reduced depth. It never skips an applicable lane silently.

So launch-all alone gives a real, shallower review. Install --all, or the specialists you care about most (privacy, consent, accessibility, admin), when depth matters.

What is in it

40 launch checks in five families, plus 12 conditional compliance domains, each owned by a named skill. A check that does not apply is recorded as not applicable, with the reason: no marketing email means no unsubscribe work; no non-essential trackers means no cookie banner is added; no need for an FAQ means none is invented.

The full models: launch-model.md and compliance-domains.md.

The 55 skills

How the skills behave

• Facts, not suspicion. Claims are labeled observed, source-indicated, declared, inferred, unknown, or review required. "A vendor appears in source" is not "it loaded before consent"; "an unsubscribe link exists" is not "future marketing is suppressed"; "delete sets active=false" is not "the data is deleted"; "no horizontal scrollbar" is not "works on a phone". Unknown is never quietly turned into a pass or a failure.
• Your design, not ours. Anything that creates or changes visible UI (404 pages, consent controls, legal pages, forms, an FAQ, an admin) first studies your existing design system, then reuses your components, and never imposes a ReadyVibe look or adds a new UI library.
• Built for your site, not from a template. admin-dashboard reads your data model and features, works out who operates your product and what they do, and builds a full admin whose navigation, vocabulary, and screens are yours, using only real data: never invented revenue, charts, or users.
• Legal specifics are looked up, not remembered. No skill carries a legal database. When a rule matters, the skill reads the current text at an official source, cites it, separates the source from its interpretation, and marks applicability review required. If it cannot look it up, the answer is unknown. Starting points are in official-sources.md.
• It will not invent. No made-up legal terms, addresses, retention periods, testimonials, FAQ answers, or metrics.
• It says when to stop. Regulated domains, exposed credentials, and genuinely unresolved legal facts are escalated to a human, in proportion to the real risk.

Bundled helpers

Where an agent cannot reliably do a job by reading files, a skill carries a small zero-dependency Node script (Node 22 or newer) in its own folder. None call a model, and none need anything from this repository.

A helper that reads nothing (server down, wrong URL) reports SITE_NOT_READ and exits non-zero, never a clean result. observe-runtime refuses to submit forms on non-local origins unless you explicitly allow it.

Marketplace bundle (Agensi)

For marketplaces that detect a single SKILL.md, the repository builds one extra artifact, dist/agensi/ReadyVibe-Skills.zip: the whole collection as one skill named ReadyVibe Skills, with a master router and all 55 methods, helpers, and references inside. It is built with pnpm marketplace:bundle (validate only: pnpm marketplace:bundle:check) and is not committed. It does not change npx skills add. See docs/marketplace-distribution.md.

What ReadyVibe does not promise

It is not a lawyer, a WCAG conformance audit, a penetration test, or a promise that a search engine will index a page. It will not say "GDPR compliant", "CCPA compliant", "fully compliant", "accessible", or "secure" as an unqualified result. Whether an agent following these skills does good work on your project depends on the agent and your project, so review what it changes. Skills are instructions your agent runs with your project's permissions: read a skill before you install it.

Repository

skills/<category>/<name>/   the skills (each folder is self-contained once installed)
scripts/                    canonical helper scripts
docs/                       current-model.md (source of truth), references/, skill-layout.md, release docs, archive/
tools/                      sync-skills.mjs, lint-skills.mjs, verify-public-install.mjs
tests/                      helper, lint, and installability tests
fixtures/                   synthetic sites and planted-defect projects

Start with docs/current-model.md. Earlier CLI/engine platform designs are kept in docs/archive/ as historical only.

Contributing, security, and releases

• CONTRIBUTING.md: skill layout, the frontmatter contract, companions, and how to run the checks
• SECURITY.md: how to report a vulnerability
• CHANGELOG.md: what changed in each release. Versions follow SemVer.

License

Apache-2.0. See LICENSE and NOTICE. Laws and standards are cited from their official sources, not copied or relicensed.