wondelai/skills: 51 book frameworks and 14 guided journeys packaged as SKILL.md files
Wondel.ai Agent Skills — Business, Marketing, UX & Coding Frameworks from Bestselling Books. 50 skills + 12 guided journeys for Claude Code, Codex, Cursor & other agentskills.io agents.
At a glance
- What is it?
- A Shell repository that turns business, marketing, UX and engineering books into installable agent skills, plus metaskills that keep their state in your project's docs/ folder. The value is in the packaging and the journey state, not in the frameworks themselves.
- Who is it for?
- Adopt it if your agents already run against a real codebase or a real product decision and you want the framework to arrive with the task instead of in a separate tab. Skip it if you need a single deep tool for one discipline, because you will be installing 65 skills to use a handful, and the repository does not document per-skill depth.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 5 days ago.
- What is it written in?
- Mainly Shell, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
The problem is framework recall, not framework supply
Every engineering team already knows that Jobs to Be Done, the Mom Test and Clean Architecture exist. The failure mode is that the framework is not present at the moment the work happens. Someone reads the book, applies it for a sprint, and the next feature ships with the same unvalidated assumptions as the last one. wondelai/skills addresses that gap by shipping the framework as an agent skill: a SKILL.md file that the agent loads when the task matches, so the questions arrive inside the conversation instead of in a separate document nobody opens. The target user is a developer or product person running Claude Code, Codex, Cursor or another agentskills.io-compatible agent who wants the method attached to the work. It is not aimed at someone who wants to read about the frameworks. The repository assumes familiarity and delivers procedure.
Two layers: 51 skills and 14 metaskills that orchestrate them
The repository describes 65 skills in total, split into 51 expert frameworks drawn from bestselling books and industry style guides, and 14 metaskills. The split matters because the two layers behave differently. A skill such as refactoring-ui or crossing-the-chasm is a single body of method. A metaskill such as create-business, improve-code-quality or conversion-optimization is a guided journey that orchestrates the skills underneath it step by step. The README states that each metaskill asks you the decision questions phase by phase and keeps its state in your project's docs/ folder, which is the mechanism that lets a journey survive across sessions. That state file is the substantive design choice here. Without it, a multi-phase journey would restart from zero on every context reset. With it, the agent has a place to read what phase you reached and what you already decided. The 14 journeys cover creating, improving and growing a business, website or app, plus code quality, technical debt, architecture, conversion optimization and general optimization.
Installation paths: plugin marketplace, skills.sh, Codex plugins
There are three documented routes. The first is the Claude Code plugin marketplace, added with /plugin marketplace add wondelai/skills and then installed by collection, for example /plugin install product-strategy@wondelai-skills for Jobs to Be Done, Negotiation and the Mom Test, or /plugin install code-craftsmanship@wondelai-skills for Clean Code, Refactoring Patterns, Software Design Philosophy, Pragmatic Programmer, DDD, Working with Legacy Code and the Google Docs Style guide. The second is skills.sh, where npx skills add wondelai/skills --all --global installs everything, and npx skills add wondelai/skills/jobs-to-be-done --global installs one skill. The third targets Codex, which the README says adopted the same open SKILL.md standard. Individual skills install with the same npx form, and the repository ships a Codex plugin marketplace at .agents/plugins/marketplace.json containing the same 10 collections, which Codex auto-discovers in a clone. The collection names are the practical unit of choice: product-strategy, ux-design, marketing-cro, sales-influence, product-innovation, strategy-growth, team-motivation, code-craftsmanship, systems-architecture and metaskills.
Where the packaging stops being an advantage
The collection structure is also the main weakness. Installing product-strategy gives you three frameworks whether you wanted all three or not, and installing the metaskills collection pulls in 14 journeys that each orchestrate skills from other collections. The README does not describe what happens when a metaskill references a skill you did not install, and it does not document the size or depth of any individual SKILL.md. A skill can be a short checklist or a long procedure, and nothing in the supplied material distinguishes the two. That is the thing to check before committing. There is a second constraint: the state lives in docs/ inside your project, which means the journey is only as durable as that folder. If docs/ is gitignored, cleaned by a build step, or excluded from the agent's context, the cross-session continuity the README promises does not hold. The repository is written in Shell, which tells you the packaging is scripted, but it says nothing about runtime dependencies beyond the agent itself.
Compared with a single-purpose skill collection
The obvious alternative is a narrower collection: something like Anthropic's own skill examples or a repo that only covers code review. The difference is scope versus depth. A narrow collection can afford to make each skill long and specific, because it only has to be good at one thing. wondelai/skills bets on breadth across business, marketing, UX and engineering, and the metaskills are the mechanism that makes breadth usable: instead of choosing between 51 frameworks, you start a journey and let it pull the relevant ones in sequence. That bet only pays off if the journeys are well sequenced. The README describes the journeys as asking decision questions phase by phase, which is a claim about structure, not about quality. You cannot verify the sequencing from the README alone. A team that already has a working review process for code and a separate process for product discovery will get less from the breadth than a small team that has neither.
Release cadence and what it implies for maintenance
The repository is not archived and the last push is dated 2026-08-29, the same day as the v1.8.0 release, which added a technical-documentation skill based on the Google developer docs style guide. v1.7.0 added conversion-optimization and architecture-optimization. v1.6.0 added support for Agent Plugins, described in the README as spec-conformant packaging at agent-plugins.org. That is roughly one feature release every two to three weeks across the visible window, and the pattern is additive: new skills and new packaging formats, not rewrites of existing ones. The practical cost is re-installation. If you installed individual skills with npx skills add, a new release does not update them for you, and the README does not document an upgrade command. If you installed via the plugin marketplace, the collection you subscribed to will pick up additions to that collection. The licence is MIT, which permits commercial use and modification, but this is a description of the licence text and not legal advice; check the LICENSE file in the repository for the terms that actually bind you.
Who should install this, and what to read first
Install it if you run an agentskills.io-compatible agent against real work and you want method to arrive with the task. Start with a single skill rather than the full set: npx skills add wondelai/skills/jobs-to-be-done --global is the cheapest way to see whether the format suits you, and the README lists jobs-to-be-done, cro-methodology and refactoring-ui among the individually installable skills. Do not install it if you need one discipline done deeply, because the breadth works against you, and do not install the metaskills collection until you have confirmed that the skills it orchestrates are present in your environment. The thing to verify first is the content of one SKILL.md file. Open it and check whether it contains phase-by-phase decision questions or a compressed summary of the book. The README promises the former for metaskills and says nothing about the 51 individual skills, so that distinction is the one that determines whether this repository changes how your team works or just adds context to every prompt.
Editorial conclusion
Adopt it if your agents already run against a real codebase or a real product decision and you want the framework to arrive with the task instead of in a separate tab. Skip it if you need a single deep tool for one discipline, because you will be installing 65 skills to use a handful, and the repository does not document per-skill depth. Before installing, open the SKILL.md for the one skill you actually intend to use and confirm it contains decision steps rather than a summary of the book.
Community notes