By 23min
Repo-agnostic engineering rituals: branching (patch/fix/chore), TDD cycle, code review, doc hygiene. Speaks the language of the work (milestone, AC, branch, test, PR) without coupling to any specific planning kernel.
Mechanical checks over the project's narrative documentation tree (`docs/`). Reports broken code references, removed-feature docs, orphan files, and documentation TODOs. Reports only — never rewrites prose. Use before opening a PR that touches code with documentation impact, or as a periodic doc-hygiene pass.
One-off branch-and-PR ritual for fixes, chores, or tweaks too small to warrant a milestone — typos, config nudges, single-line bug fixes, dependency bumps. Creates a fix/patch/chore branch, lands one focused change, gates commit and push behind explicit human approval. Use when the user describes a single focused change tied to one issue (or none).
Code review checklist with verdict (approve / request-changes / questions). Reviews a diff for correctness, AC coverage, branch-coverage discipline, conventions, and documentation hygiene. Findings are classified blocking vs. non-blocking with file:line references. Use when the user says "review this", "check my changes", or before a PR is opened.
Red/green/refactor for a single acceptance criterion or feature unit, ending with a hard-rule branch-coverage audit. Write a failing test, write the minimum code to pass, refactor, then walk every reachable conditional branch and confirm an explicit test exercises it. Use during milestone implementation and inside `wf-patch` when the change touches logic.
A Claude Code plugin marketplace shipping engineering-ritual skills and role agents that complement aiwf.
Pre-alpha. Marketplace and plugin manifests scaffolded; no skills shipped yet.
aiwf is a narrow planning kernel: six entity kinds (epic, milestone, ADR, gap, decision, contract), stable IDs, validators, git-trailer history. It deliberately has no opinion about how the work gets done — branching, TDD, code review, milestone lifecycle, release ceremonies.
This repo ships those opinions, separately, as Claude Code plugins. Two plugins, both optional:
| Plugin | Contains | When to install |
|---|---|---|
| aiwf-extensions | Milestone-lifecycle skills (aiwfx-start-milestone, aiwfx-wrap-milestone, aiwfx-wrap-epic, aiwfx-release, aiwfx-track, aiwfx-record-decision, …) plus four role agents (planner, builder, reviewer, deployer) | If you use aiwf as your planning kernel and want the milestone-driven engineering loop |
| wf-rituals | Generic engineering rituals (wf-patch, wf-tdd-cycle, wf-review-code, wf-doc-lint) | If you want the engineering rituals on their own — works on any repo, with or without aiwf |
In a Claude Code session, from inside a consumer repo:
/plugin marketplace add 23min/ai-workflow-rituals
/plugin install aiwf-extensions@ai-workflow-rituals
/plugin install wf-rituals@ai-workflow-rituals # optional
Updates: /plugin update <name>@ai-workflow-rituals. Auto-update on startup is enabled by default while the marketplace is in active development.
Claude Code asks where to install. Pick Project scope by default. It writes .claude/settings.json into the consumer repo, which means:
User scope is only right if you're sure you want these plugins available in every project on your machine. Local scope (gitignored) is right for short-term experimentation in a single repo.
For the planning kernel itself, see ai-workflow-v2:
go install github.com/23min/ai-workflow-v2/tools/cmd/aiwf@latest
aiwf init # in your consumer repo
The two plugins differ in how tightly they couple to aiwf. This is deliberate.
aiwfx-* skills — call aiwf shell verbs, reference aiwf's ID format (E-NN, M-NNN, …), use aiwf's status vocabulary. By definition coupled. If you don't use aiwf, this plugin is dead weight.wf-* skills — speak the language of the work (milestone, acceptance criteria, branch, test, PR) without mentioning aiwf, its IDs, its status vocabulary, or its directory layout. By design they survive an aiwf swap. They work the same against a repo using aiwf, against a repo using a different planning kernel, or against a repo using none.This is a discipline, not an enforcement. Every wf-* skill goes through a coupling-boundary review at port time: "if aiwf were deleted tomorrow, would this skill still make sense?" If no, it belongs in aiwfx-*.
ai-workflow-rituals/
├── .claude-plugin/
│ └── marketplace.json
├── plugins/
│ ├── aiwf-extensions/
│ │ ├── .claude-plugin/plugin.json
│ │ ├── skills/aiwfx-*/SKILL.md
│ │ ├── agents/{planner,builder,reviewer,deployer}.md
│ │ └── templates/
│ └── wf-rituals/
│ ├── .claude-plugin/plugin.json
│ ├── skills/wf-*/SKILL.md
│ └── templates/
├── README.md
└── CHANGELOG.md
Each plugin's directory layout follows Claude Code's plugin reference. Contents are auto-discovered.
| Plugin | Stage |
|---|---|
| Marketplace manifest | scaffolded |
aiwf-extensions plugin manifest | scaffolded |
wf-rituals plugin manifest | scaffolded |
aiwf-extensions skills | not started |
aiwf-extensions agents | not started |
aiwf-extensions templates | not started |
wf-rituals skills | not started |
wf-rituals templates | not started |
The porting plan and the rationale for what's in scope vs. deferred lives in ai-workflow-v2/docs/rituals-plugin-plan.md.
Apache-2.0. See LICENSE.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub 23min/ai-workflow-rituals --plugin wf-ritualsMilestone-lifecycle skills and role agents for aiwf. Wraps aiwf's planning verbs with opinionated rituals (TDD-driven milestone start, contract-checked milestone wrap, release tagging, in-flow ADR/decision recording, tracking-doc convention).
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
Develop, test, build, and deploy Godot 4.x games with Claude Code. Includes GdUnit4 testing, web/desktop exports, CI/CD pipelines, and deployment to Vercel/GitHub Pages/itch.io.
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Comprehensive PR review agents specializing in comments, tests, error handling, type design, code quality, and code simplification
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
v9.44.1 — Patch release for Gemini environment/version detection and qwen auth gating. Run /octo:setup.