Anthropic Released Skills on October 16. We Started Building the Same Day.
When Anthropic launched Skills for Claude on October 16, 2025, most of the coverage focused on what they are in the abstract. A new way to customize Claude. Reusable instructions. Blah blah blah. That is the marketing version.
Here is the practical version: Skills let you turn Claude into a specialist for a specific task. Instead of writing the same detailed prompt every time you need a contract reviewed, or a client onboarded, or a weekly report generated, you define the instructions once as a skill. Then you -- or anyone on your team -- can invoke that skill with a single command. Same quality. Same format. Every time.
We saw the announcement and immediately understood what this meant for client work. Within four days, we had built four production skills that are now part of our standard workflow. This post walks through each one, explains what it does, and shows how much time it saves. No theory. Just the work.
What Skills Actually Are (Plain English)
A skill is a set of instructions that tells Claude how to handle a specific type of task. Think of it as a playbook. You define the steps, the format, the criteria, and the edge cases once. Then Claude follows that playbook every time the skill is invoked.
In Claude Code, a skill is a markdown file that lives in your project. The file describes the task, the inputs it expects, the steps to follow, and the format for the output. When you tell Claude to use that skill, it reads the file and follows the instructions. That is it. No API configuration, no complex setup. Write a markdown file, tell Claude to use it.
The power is in consistency and delegation. Before skills, our team members would each prompt Claude slightly differently for the same task. The output quality varied. With skills, the instructions are standardized. A junior team member invoking the contract review skill gets the same thorough analysis that a senior consultant would get. The expertise is encoded in the skill, not in the person using it.
A skill is expertise encoded as instructions. A junior team member invoking a well-built skill gets the same output quality as a senior consultant. That changes how you scale a team.
Skill 1: Contract Review
This was the first skill we built because contract review is one of our most frequent client requests and the one where consistency matters most.
The skill file defines a 14-point checklist that Claude follows for every contract it reviews. The checklist covers: party identification and authority, scope of work clarity, payment terms and late penalties, termination clauses, liability caps, indemnification language, non-compete and non-solicitation provisions, intellectual property ownership, confidentiality obligations, force majeure provisions, dispute resolution mechanisms, governing law, amendment procedures, and automatic renewal traps.
For each item, the skill specifies what "good" looks like, what common problems to flag, and how to rate the risk level (low, medium, high). The output is a structured report with a summary at the top, detailed findings for each checklist item, and a recommended actions section at the bottom.
Before this skill, contract review took 45-60 minutes per document because whoever was doing it had to remember the full checklist, decide on formatting, and maintain consistency. With the skill, it takes about 8 minutes -- most of that is reading the output and verifying the flagged items. The quality is more consistent than our manual reviews because the skill never skips a checklist item, which humans occasionally do when they are tired or rushed.
Time saved per contract: roughly 40 minutes. We review 15-20 contracts per month. That is 10-13 hours reclaimed every month from a single skill.
Skill 2: Client Onboarding Welcome Docs
When a new client signs with us, we generate a welcome package: an overview of our process, key contacts, communication preferences, timeline expectations, access and credential setup instructions, and a custom project brief based on their discovery call notes.
This used to take 90 minutes per client because every package was manually assembled from templates and customized. The templates were in Google Docs, the customization was done by whoever happened to be handling onboarding that week, and the quality was -- let's be honest -- inconsistent.
The onboarding skill takes the client's name, industry, project scope, and discovery call notes as inputs. It generates the full welcome package in our standard format, customized with industry-specific context and project-relevant details. The skill knows our process because we encoded it: week 1 is discovery and audit, week 2 is strategy, weeks 3-4 are build and iterate, and ongoing is optimization and support.
The output includes a personalized welcome letter, a project timeline with milestones, a communication guide (who to contact for what, response time expectations), an access checklist for tools and platforms they will need, and a FAQ section that addresses the three most common questions for their industry.
Time saved per onboarding: about 70 minutes. We onboard 4-6 new clients per month. That is 5-7 hours reclaimed, and every client now gets the same thorough welcome experience regardless of who handles the onboarding.
Skill 3: Weekly Reporting
Every Monday, we send each active client a progress report. The report covers: work completed in the previous week, metrics and outcomes (where measurable), work planned for the coming week, blockers or decisions needed from the client, and budget utilization.
Before the skill, assembling these reports meant pulling data from our project management tool, checking analytics dashboards, reviewing Slack threads for context, and writing it all up in a consistent format. Each report took 20-30 minutes. With 12-15 active clients, that was an entire Monday morning consumed by report writing.
The weekly reporting skill takes structured inputs -- completed tasks, metrics data, planned tasks, and any blockers -- and formats them into our standard report template. It writes the narrative sections that connect the data points, adds context about why certain metrics moved, and flags items that need client attention. The skill also knows our formatting standards: bullet points for completed work, a metrics table for numbers, and bold text for action items.
Time saved per report: about 15 minutes. Across 12-15 clients per week, that is 3-4 hours every Monday that our team gets back for actual client work instead of report formatting.
Each of these skills took less than an hour to build. Combined, they save us roughly 20 hours per month. The ROI on that first afternoon of skill-building has already paid for itself many times over.
Skill 4: Proposal Writing
Our proposals follow a specific structure that we have refined over 50+ engagements: executive summary, problem statement (in the client's language, not ours), proposed approach, deliverables and timeline, investment and payment terms, team and qualifications, and next steps.
The proposal skill takes the prospect's company name, industry, pain points from the discovery call, proposed scope, and budget range. It generates a complete first draft in our format, using our tone (direct, no jargon, specific about outcomes rather than vague about capabilities).
The skill also includes instructions for what not to do: do not overpromise timelines, do not use the word "synergy," do not include generic case studies that are not relevant to the prospect's industry, and do not bury the price at the end. Our proposals lead with the investment section because we have found that transparency about cost builds trust faster than hiding it behind 10 pages of capability descriptions.
Before the skill, proposals took 2-3 hours to write. The skill produces a first draft in about 5 minutes that requires 30-40 minutes of editing and customization. The total time dropped from roughly 2.5 hours to about 45 minutes per proposal.
Time saved per proposal: about 100 minutes. We write 6-8 proposals per month. That is 10-13 hours reclaimed.
Our Open-Source Skills Library
We open-sourced our skills library on GitHub because we think the fastest way for businesses to adopt skills is to see working examples. Reading documentation is one thing. Seeing a production skill file with real instructions is a completely different learning experience.
The library includes the four skills described above (with client-specific details removed, obviously) plus additional skills we have built for common business tasks: meeting summary formatting, competitive analysis frameworks, email drafting in different tones, and data cleaning checklists.
Each skill file includes comments explaining why specific instructions are included, so you can understand the reasoning and adapt it to your own workflows. We have found that the "why" behind each instruction is more valuable than the instruction itself when you are building your own skills.
How Any Business Can Start Building Skills
You do not need to be a developer. You do not need to be an AI expert. You need one thing: a task that you do repeatedly and that follows a pattern.
Start by picking your most repetitive task. Not your most complex one -- your most repetitive one. The task you do every week that makes you think "I should have a template for this." That is your first skill.
Write down the steps you follow when you do this task well. Be specific. Do not write "analyze the contract" -- write "check whether the liability cap is proportional to the contract value and flag anything under 2x annual fees as high risk." The more specific your instructions, the better the skill performs.
Define the output format. What should the result look like? A bulleted list? A formatted report with sections? A table? Skills work best when the output format is explicit. Do not let the model decide how to present information -- tell it exactly what you want.
Include the edge cases. What happens when information is missing? What should the skill do when the input is ambiguous? The difference between a mediocre skill and a great one is usually the edge case handling. Think about the last time this task went sideways and encode the fix into the skill.
Test it on real work. Not on hypothetical examples -- on the actual task you did last Tuesday. Compare the skill's output to what you produced manually. Where is it better? Where is it worse? Adjust the instructions and test again. Two or three iterations usually gets a skill to production quality.
The Compound Effect
Each individual skill saves a modest amount of time. The contract review skill saves 40 minutes. The onboarding skill saves 70 minutes. By themselves, these are nice-to-haves.
But skills compound. Four skills saving 20 hours per month is just the beginning. We are building new skills almost every week now as we identify patterns in our work. By the end of the quarter, we expect to have 12-15 production skills covering the majority of our repeatable workflows.
The more interesting compounding effect is qualitative, not quantitative. When you encode your best practices into skills, the floor of your output quality rises. The worst version of your team's work gets better because the skill enforces the standard. That is harder to measure than time saved, but it might matter more.
Skills are not going to replace judgment, expertise, or client relationships. They are going to replace the mechanical parts of knowledge work -- the formatting, the checklists, the templates, the first drafts. That frees up your time for the parts of the work that actually require a human: understanding context, making judgment calls, building trust, and solving problems that do not have playbooks yet.
If you are not building skills yet, start this week. Pick one task. Write one skill file. Test it on real work. You will wonder why you ever did it the old way.
Since writing this post, we eventually open-sourced our entire skills library -- over 100 production-ready skills spanning sales, operations, content, development, and more. Free on GitHub. We also built our interactive AI Tool Belt where you can compare and test AI tools side by side. Between the skills library, the AI Tool Belt, and Vibe Quest, we have built more free resources than most consulting firms charge for -- because the fastest way to grow the ecosystem is to make the on-ramp as smooth as possible.
Skills are not going to replace judgment, expertise, or client relationships. They are going to replace the mechanical parts of knowledge work -- the formatting, the checklists, the first drafts. That frees up your time for the parts that actually require a human.