The Ready Suite

Eleven composable AI skills covering the full arc from idea to launch. Implements the Agent Skills standard. Plug them into Claude Code, Codex, Cursor, Windsurf, pi, OpenClaw, or any harness that parses SKILL.md frontmatter natively. Each skill stands alone. They compose by reading and writing well-known artifact paths, never by calling each other.

Successor available: aihxp/arc-ready. The same eleven-tier discipline, every named failure mode preserved, every grep test preserved, every artifact path unchanged, consolidated into a single skill and a single repo. One install replaces eleven. The eleven-skill suite remains available and supported; arc-ready is the recommended starting point for new projects. See arc-ready/MIGRATION.md for the migration matrix.

This is the monorepo. Every skill lives under skills/<skill-name>/ with its own SKILL.md, references library, and CHANGELOG. The byte-identical SUITE.md ships at the hub root and inside every skill subdirectory, so the suite map is visible from any entry point.

The eleven skills, four tiers

ORCHESTRATION TIER       PLANNING TIER             BUILDING TIER         SHIPPING TIER
------------------       ------------------        ----------------      ------------------
kickoff-ready  ->        prd-ready           ->    repo-ready      ->    deploy-ready
                         architecture-ready  ->    production-ready ->   observe-ready
                         roadmap-ready       ->                          launch-ready
                         stack-ready         ->                          harden-ready

What each skill owns

Skill	Tier	One-line purpose	Version	Path
kickoff-ready	Orchestration	Sequence the ten specialists for a greenfield project from raw user intent.	1.1.8	skills/kickoff-ready
prd-ready	Planning	Define what we're building and for whom.	1.0.16	skills/prd-ready
architecture-ready	Planning	Design how the big pieces fit together.	1.0.15	skills/architecture-ready
roadmap-ready	Planning	Sequence work over time.	1.0.14	skills/roadmap-ready
stack-ready	Planning	Pick the right tools for the job.	1.1.21	skills/stack-ready
repo-ready	Building	Set up the repo with production-grade hygiene.	1.7.0	skills/repo-ready
production-ready	Building	Build the app to production grade.	2.6.8	skills/production-ready
deploy-ready	Shipping	Ship the app to real environments.	1.0.20	skills/deploy-ready
observe-ready	Shipping	Keep the app healthy once it's live.	1.0.19	skills/observe-ready
launch-ready	Shipping	Tell the world the product exists.	1.0.17	skills/launch-ready
harden-ready	Shipping	Survive adversarial attention; prove it to an auditor.	1.0.13	skills/harden-ready

See it end-to-end: ready-suite-example

A complete dogfood of the suite applied to a fictional B2B SaaS product (Pulse, a Customer Success ops platform) lives at examples/ready-suite-example/. All eleven artifacts on disk at canonical .{skill}-ready/ paths, with cross-references intact, plus an AGENTS.md (kickoff-ready emit), a DESIGN.md (Google Labs format, scaffolded by production-ready), and a DOGFOOD.md reflection on what the suite caught, what it missed, and what would change in a v1.x of the suite.

The product, the people, and the dates are fictional; the artifacts are rigorous (each passes its own skill's grep tests for named failure modes). Use the dogfood to see what the suite produces end-to-end before installing it on a real project.

Why eleven skills, not one

Generic AI coding assistants do everything badly. The ready-suite splits the work into eleven skills with tight scope and distinct trigger surfaces, so the harness can route precisely and each skill can refuse work that belongs to a sibling.

A skill at this scope can be opinionated. It can carry concrete have-nots, named failure modes, and acceptance gates per step. It can ship references with worked examples and per-domain considerations. It can stay current without bloating, because each release only touches one tier of the lifecycle.

A single mega-skill cannot do any of these things at the same time without becoming the "god skill" the suite exists to prevent.

Standards compliance

Ready-suite skills implement the Agent Skills standard. The contract is plain: a SKILL.md with YAML frontmatter and an optional references/ directory the harness loads on demand. Any harness that parses this format runs every skill first-class. Verified harnesses today: Claude Code, Codex, Cursor, Windsurf, pi, OpenClaw. pi and OpenClaw both load skills from the neutral ~/.agents/skills/ path defined by the standard, so future AgentSkills-compatible harnesses inherit support with no per-tool integration work.

The suite also meets the AGENTS.md cross-tool agent-brief standard (governed by the Linux Foundation's Agentic AI Foundation), the project-root brief read natively by Codex CLI, GitHub Copilot, Cursor, Windsurf, Aider, Zed, Warp, Roo Code, Jules, Factory, Amp, Devin, and others. kickoff-ready emits a project-root AGENTS.md mapping the suite's artifact paths when none exists (Step 6 sub-step 6a); repo-ready scaffolds AGENTS.md as the canonical agent brief with CLAUDE.md as a thin overlay or symlink. Both respect any user-authored AGENTS.md already on disk.

The suite consumes the Google Labs DESIGN.md format (Apache 2.0; YAML frontmatter holds machine-readable design tokens, the markdown body holds rationale plus an Agent Prompt Guide). production-ready detects a project-root DESIGN.md in Step 3 sub-step 3a and consumes it via DTCG export, Tailwind v4 export, or direct YAML read; the linter (npx @google/design.md lint) runs before component code so WCAG and token-resolution failures block the slice deterministically. When no DESIGN.md is present, production-ready falls back to its archetype + 5-decision derivation and optionally scaffolds a DESIGN.md from the chosen tokens.

Claude Code plugin install

Inside Claude Code:

/plugin marketplace add aihxp/ready-suite
/plugin install ready-suite@ready-suite

This adds the suite's marketplace and installs the ready-suite meta plugin, which depends on every specialist (prd-ready, architecture-ready, ..., harden-ready). One install command, eleven skills.

Want only one skill? Install it directly:

/plugin install prd-ready@ready-suite

The marketplace lives in this hub repo at .claude-plugin/marketplace.json. Each plugin is vendored under plugins/<skill-name>/ with its own manifest. See references/PLUGIN-RESEARCH.md for the format research that shaped these decisions.

Plugin maintainers refresh the vendored copies from the canonical content under skills/<skill-name>/ whenever a sibling's version bumps, then commit and tag plugin-v<x.y.z>.

One-command install (Codex, Cursor, pi, OpenClaw, all platforms)

Installs every skill into every detected harness (Claude Code, Codex, Cursor, plus the neutral Agent Skills path read by pi and OpenClaw) with a single shell command. Idempotent. Re-run anytime.

git clone https://github.com/aihxp/ready-suite.git ~/Projects/ready-suite
bash ~/Projects/ready-suite/install.sh

What it does:

1. Detects which harnesses you have (~/.claude, ~/.codex, ~/.cursor, ~/.pi, ~/.openclaw). Skips ones that aren't installed.
2. When pi or OpenClaw is detected, also writes to the neutral ~/.agents/skills/ path defined by the Agent Skills standard. Both harnesses read this path natively, and any future AgentSkills-compatible harness inherits support for free.
3. For each of the eleven skills, symlinks SKILL.md and references/ from skills/<skill>/ into every detected harness's skills directory.
4. Existing non-symlink installs are backed up to <target>.backup-<timestamp>/ first.

Edit a file in skills/<skill>/ (or git pull to update) and the change is live in every harness immediately. No re-install needed.

Verbose mode shows every step:

bash install.sh -v

Per-skill install (manual)

The eleven skills live under skills/<skill-name>/ in this monorepo. To install just one skill into a single harness, copy or symlink that skill's SKILL.md and references/ into the harness's skills directory:

# Claude Code (example: just prd-ready)
ln -s ~/Projects/ready-suite/skills/prd-ready/SKILL.md ~/.claude/skills/prd-ready/SKILL.md
ln -s ~/Projects/ready-suite/skills/prd-ready/references ~/.claude/skills/prd-ready/references

For most users the hub installer (bash install.sh) is simpler and idempotent. It already does the symlinking for every detected harness.

Windsurf or other agents: point your project rules or system prompt at the skill's SKILL.md (skills/<skill-name>/SKILL.md) and load reference files as needed.

Uninstall

Removes the suite's symlinks from every detected harness. Leaves the in-tree skills under skills/ and any .backup-*/ directories untouched.

bash uninstall.sh

Composition

Skills do not call each other. The harness is the router.

• Plain harnesses (Claude Code, Codex, Cursor, Windsurf): the harness's skill router invokes whichever skill matches the user's words.
• Orchestrators (GSD, BMAD, Spec Kit, Superpowers, custom): see ORCHESTRATORS.md for command-by-command integration patterns. The orchestrator knows about the suite; the suite stays orchestrator-agnostic.
• Artifacts are the contract. Every skill produces files at .{skill}-ready/*.md paths. Downstream skills read them. Skills compose without invoking each other.
• Routing ambiguity: when a user phrase plausibly matches more than one skill ("set up CI," "GitHub Actions," "make this production-ready"), see references/TRIGGER-DISAMBIGUATION.md for the canonical resolution table.

Composition principles

1. Tight scope beats combined scope. Splitting is the first move when scope drifts.
2. The harness is the router. No skill calls another.
3. Artifacts are the contract. Skills compose by file paths, not function calls.
4. Version tracking is mandatory. Every skill carries version, updated, suite, tier, upstream, and downstream in frontmatter.
5. No skill duplicates another. If work overlaps, one skill owns the canonical content and siblings reference it.
6. Graceful degradation. Every skill works standalone. Upstream absence is handled with sensible defaults, not errors.
7. Byte-identical SUITE.md across siblings. When any sibling releases, every sibling's SUITE.md row bumps. The known-good-versions table is authoritative from any entry point.

Versions

The SUITE.md known-good-versions table at the hub root is byte-identical with every skills/<skill>/SUITE.md. To find the latest of any skill, check the table from any entry point: they always agree.

When any skill version bumps, the hub SUITE.md and every skills/<skill>/SUITE.md are updated in the same commit. The lint enforces the byte-identical invariant.

Maintenance: ready-suite-lint

The hub ships a meta-linter that mechanically enforces the suite's discipline rules: SUITE.md byte-identical between the hub and every skills/<skill>/ subdir, frontmatter version matches CHANGELOG top entry, no em-dashes / arrows / box-drawing characters in suite-authored files, and the compatible_with frontmatter declares the standards-level values.

Run locally:

bash scripts/lint.sh                   # all checks
bash scripts/lint.sh --verbose         # show ok lines
bash scripts/lint.sh suite-md-sync     # one specific check
bash scripts/lint.sh --help            # see all checks and flags

Available checks: suite-md-sync, frontmatter-version, unicode-clean, compatible-with, trigger-overlap.

The trigger-overlap check is advisory by default: it surfaces cross-skill substring overlaps in the eleven skills' description trigger phrases (e.g., "GitHub Actions" appearing in both repo-ready scaffolding triggers and deploy-ready pipeline triggers) and prompts the maintainer to verify a row exists in references/TRIGGER-DISAMBIGUATION.md. Pass --strict-triggers to make overlap warnings fail the lint.

The same lint runs in GitHub Actions on every push to main and on pull requests (.github/workflows/lint.yml).

The lint is bash 3.2 compatible (macOS default) and uses no associative arrays.

Contributing

PRs welcome. See CONTRIBUTING.md for the contribution model, the unicode rule, the bash-3.2 rule, and how to land a single-skill or coordinated cross-suite change.

For maintainers (or anyone landing a coordinated cross-suite patch), see MAINTAINING.md: the SUITE.md sync ritual, version-bump rules, and the three maintenance rituals (single-skill patch, coordinated cross-suite patch, lint regression recovery).

PR review is auto-routed via .github/CODEOWNERS. Path-specific rules name the high-blast-radius surfaces (SUITE.md, install.sh, lint, marketplace manifest); when a second maintainer is added, the CODEOWNERS file is the single place to grant review authority on those paths.

License

MIT. Each skill under skills/<skill-name>/ carries its own LICENSE file with the same terms.