We Had 100 Skills Nobody Could See
By late 2025, we had built Claude Code skills for every type of client engagement we touched. Contract review, client onboarding, proposal writing, weekly reporting, competitive analysis, meeting summaries, sales email drafting, data cleaning, code review checklists, content calendars, social media scheduling, invoice processing, lead scoring templates, SEO audits. The list kept growing.
Each skill started the same way: we would be doing a task for the third or fourth time, realize we were following the same steps, and encode those steps into a skill file. The skill file lives in the project as a markdown document -- instructions that Claude Code reads and follows consistently, every time, without forgetting steps or getting tired at 3pm on a Friday.
We wrote about the first four production skills we built in our post on how Claude Skills changed how we work with clients. But four became fourteen, and fourteen became forty, and by January 2026 we had over 100 production-ready skills sitting in a private repository. Every one of them battle-tested on real client work. Every one of them saving us measurable hours every month.
And then we had a conversation that changed our approach entirely.
We had over 100 production-ready skills sitting in a private repository. Every one of them battle-tested on real client work. And we were keeping them locked away like they were the crown jewels. That was a mistake.
The Conversation That Changed Everything
A potential client -- a mid-size marketing agency in Tampa -- called us to talk about deploying Claude across their team. They had been experimenting on their own for months. They had read our blog posts. They understood skills conceptually. But they could not build one that actually worked well.
Not because they were not smart. They were sharp operators who understood their own workflows perfectly. The problem was the gap between understanding what a skill should do and knowing how to write instructions that Claude would follow reliably. They would write a skill for client reporting, and Claude would produce a different format every time. They would write a skill for content calendars, and it would miss crucial details about platform specifications. The skills worked, sort of, but they were not production quality.
We showed them our contract review skill -- the one that follows a 14-point checklist and produces a consistent risk-rated report every time. They looked at it for about thirty seconds and said, "Oh. That is what we are doing wrong. We did not know how specific the instructions needed to be."
That moment crystallized something. Our skills were not valuable because the tasks they automated were unique. Contract review, client onboarding, proposal writing -- every consulting firm does these things. Our skills were valuable because we had figured out how to write instructions that Claude follows consistently. The expertise was in the instruction design, not in the existence of the instructions.
So we open-sourced all of them.
What Is in the Library
The OneWave Claude Skills Library on GitHub contains over 100 production-ready skills organized across six categories.
Sales and Business Development: lead qualification frameworks, prospect research templates, cold email drafters (with tone controls that actually work), proposal generators, competitive battle cards, and pipeline analysis skills. These are the skills our sales consulting clients use most -- the lead qualification skill alone saves our clients an average of 8 hours per week.
Business Operations: contract review (the 14-point checklist version), client onboarding document generators, weekly and monthly reporting skills, meeting summary formatters, project status dashboards, and invoice processing workflows. These are the backbone skills -- the ones every business needs regardless of industry.
Content Creation: blog post outliners and drafters, social media content calendars, email newsletter generators, case study frameworks, press release templates, and SEO content audit skills. Each one includes platform-specific formatting instructions -- the LinkedIn skill knows about character limits and hashtag strategy, the newsletter skill knows about subject line optimization.
Development: code review checklists, bug report templates, release notes generators, technical documentation skills, API documentation formatters, and migration planning skills. These are the skills our dev team uses daily -- built from the patterns we see across dozens of client codebases.
Data and Analysis: data cleaning checklists, competitive landscape analysis frameworks, market research templates, financial modeling summaries, customer feedback categorization, and survey analysis skills. These encode the analytical frameworks we have developed over 50+ engagements.
Customer Success: health score assessment skills, churn risk identification frameworks, QBR preparation templates, customer feedback synthesis, and escalation response drafters. These are newer additions, born from our expanding work with customer success teams.
Why Open Source
Let me be direct about the business reasoning because I think transparency here is more interesting than false modesty.
Reason 1: Skills are only valuable if people know how to use them. A skill file sitting in a GitHub repository does not help anyone who has never used Claude Code. But a skill file sitting in a GitHub repository that someone discovers, downloads, tries, and then thinks "I need 20 more of these customized for my specific business" -- that person calls us. The library is the best demonstration of our capabilities we have ever created. Better than any case study, any sales deck, any webinar.
Reason 2: Our competitive advantage is not the skills themselves. This is the key insight that made the decision easy. Anyone can look at our contract review skill and copy the structure. What they cannot copy is the knowledge of which skills a mid-size law firm needs versus a marketing agency versus a logistics company. They cannot copy the experience of deploying 100+ skills across 50+ client engagements and knowing exactly which ones deliver ROI in week one versus which ones need three months of customization. The skills are the product catalog. The expertise is knowing which products to deploy, in what order, customized in what way. That expertise comes from doing the work, not from reading a repo.
Reason 3: Growing the ecosystem benefits everyone, including us. The bigger the Claude Code ecosystem gets, the more businesses adopt it, the more demand exists for expert help implementing it. We are betting that growing the pie is more valuable than guarding our slice. Every person who learns to use skills from our library is one more person who understands the value of AI workflow automation. Some percentage of those people will need custom work that goes beyond what a public library provides. We want to be the team they think of first.
We have gotten more inbound leads from the skills library than from any paid marketing we have ever done. The irony of giving things away generating more business than selling things is not lost on us.
How People Are Using Them
The response has been more enthusiastic than we projected. The repository has been forked hundreds of times in its first two months. The community contributions are the most interesting part -- people are submitting skills we never would have thought to build. Someone submitted an entire suite of skills for real estate transaction management. Another user contributed a set of skills for academic research workflow automation. A freelance designer submitted skills for client brief processing and project scoping.
The most popular skills, based on issue threads and feedback, are not the flashy ones. The contract review skill gets the most attention because it solves a universal problem. The weekly reporting skill is second because every team that uses it reclaims 3-4 hours per week immediately. The proposal writing skill is third because the time savings are dramatic -- from 2.5 hours per proposal to about 45 minutes.
What surprised us was how many people use the library as a learning resource rather than a tool library. They download the skills not to use them directly, but to study the instruction patterns and then build their own. That is exactly the behavior we hoped for. The skill files include detailed comments explaining why each instruction is structured the way it is -- what makes Claude follow the instructions consistently versus what causes it to drift. Those comments are arguably more valuable than the skills themselves.
The Business Case for Giving Things Away
I know what the skeptics are thinking. "You just gave away your entire skill library. How is that a business strategy?" Here is how.
Before the library, our sales process was: cold outreach, discovery call, proposal, close. Conversion rate from discovery call to close was around 35%. Decent for consulting.
After the library, the sales process looks different. Someone finds the library on GitHub or through a blog post. They download a few skills. They try them. They realize skills are genuinely useful. They hit a limitation -- the generic skill does not fit their specific workflow, or they need skills that do not exist yet, or they want someone to deploy skills across their entire team and train everyone. They reach out to us. When they show up for a discovery call, they already understand what we do, they have already experienced the value, and they are asking "how much to customize this for us" instead of "convince me this works."
Conversion rate from discovery call to close for library-sourced leads: 62%. Nearly double our cold outreach conversion. The reason is obvious -- these prospects have already sold themselves. They experienced the product before they talked to a salesperson. That is a fundamentally different conversation.
We have gotten more inbound leads from the skills library than from any paid marketing we have ever done. More than LinkedIn ads, more than Google ads, more than sponsored content. A free GitHub repository outperformed our entire paid acquisition budget. The irony is thick and we lean into it.
How to Get Started
If you are new to Claude Code skills, here is the fastest path from zero to productive.
First, pick a skill from the library that matches something you actually do in your work. Do not browse the whole library -- find one skill that solves a real problem you had this week. The contract review skill if you review contracts. The meeting summary skill if you attend meetings that need follow-up notes. The proposal writing skill if you are about to write a proposal.
Second, read the skill file before you use it. Not just the instructions -- read the comments. They explain the reasoning behind each instruction, and understanding the reasoning is what lets you customize effectively. A skill that is 80% right for your workflow is more useful than a skill that is 100% right for someone else's workflow.
Third, customize it. Change the output format to match your existing templates. Adjust the checklist items to reflect your industry's priorities. Add instructions for edge cases you encounter in your specific work. The best skills in our library started as generic templates and became specific tools through iteration.
Fourth, contribute back. If you build a skill that works well for your industry or use case, submit a pull request. The library gets better with every contribution, and the community benefits from perspectives we do not have. We are consultants -- we know consulting workflows well. We know less about veterinary practice management or nonprofit grant writing or restaurant operations. The community fills those gaps.
What We Are Building Next
The library is not done. It will never be done -- that is the point of open source. But we have specific expansions planned.
We are building a skill composition framework -- a way to chain multiple skills together into workflows. Right now, each skill is standalone. We want to enable sequences like: run the meeting summary skill, then feed the output into the action items skill, then feed the action items into the project management update skill. Multi-step automation built from modular skill components.
We are also building industry-specific skill packs -- curated collections of skills for specific verticals. The law firm pack, the marketing agency pack, the e-commerce pack. Each pack will include the skills, a deployment guide, and a training curriculum for teams. These packs will be the bridge between the free library and our custom consulting work -- structured enough to use independently, but designed to surface the complexity that makes custom work valuable.
The library lives at github.com/OneWave-AI/claude-skills. Star it if you find it useful. Fork it if you want to experiment. Submit a PR if you build something worth sharing. And if you hit the point where generic skills are not enough and you need someone to build a custom skill architecture for your team -- you know where to find us.
Our competitive advantage was never the skills themselves. It is knowing which skills to deploy for each business, in what order, and how to customize them for maximum impact. The skills are the product catalog. The expertise is the strategy.