📊 Full opportunity report: A Skill Is A Folder, Not A Prompt: What Anthropic Learned Running Hundreds Of Them on ThorstenMeyerAI.com — validation score, market gap, and execution plan.

TL;DR

Anthropic has demonstrated that ‘skills’ for AI agents are best understood as folders containing instructions, scripts, and assets, not just prompts. This approach enhances consistency, onboarding, and institutional knowledge. The company ran hundreds of experiments internally, leading to a new framework for building and managing AI capabilities at scale.

Anthropic has revealed that its approach to building AI agent capabilities involves defining Skills as folders containing instructions, scripts, and reference assets, rather than simple prompts. This shift aims to create durable, reusable organizational assets that improve consistency and onboarding across teams.

In a detailed write-up, Anthropic described how its internal experiments with hundreds of Skills led to a new understanding: a Skill is a container, not just a prompt. Each folder can include instructions, reference documents, runnable scripts, templates, data, configuration, and hooks that activate during operation. This design allows agents to discover, read, and execute complex workflows, making organizational knowledge more durable and accessible.

The company emphasizes that Skills serve three core functions: ensuring consistent output regardless of who runs the agent, compressing onboarding by embedding tribal knowledge directly into the agent, and enabling continuous improvement through iteration. Anthropic’s internal analysis identified nine categories of Skills, ranging from library references and data analysis to process automation and infrastructure operations, with verification Skills deemed most valuable for quality control.

Technical lessons highlight that effective Skills avoid restating obvious information, focus on non-obvious, specific knowledge, and include ‘Gotchas’—traps and edge cases—captured as institutional memory. Descriptions serve as trigger definitions, matching user requests precisely to activate the correct Skill, and scripts are bundled with real code to enable automation and error handling.

At a glance
reportWhen: published recently, with ongoing intern…
The developmentAnthropic published insights from its internal experiments, showing that packaging knowledge into folder-based skills significantly improves AI agent performance and organizational consistency.
A Skill Is a Folder, Not a Prompt — Insights
AI Dispatch · Insights · 1 July 2026

A Skill is a folder, not a prompt

Anthropic published what it learned running hundreds of Skills across its own engineering org. Read as a business memo, the point is bigger than a coding trick: this is how ad-hoc prompting becomes durable institutional capability — the SOPs your agents actually follow, versioned and shared.

✕ The misconception

“A Skill is just a clever markdown prompt you save in a file.”

✓ What it actually is

A folder the agent can discover, read & run — instructions, scripts, references, templates, config & on-demand hooks.

Anatomy of a Skill — the file system is context engineering
my-skill/the unit you share & version
├─ SKILL.mdroot instructions + a description written for the model (its trigger)
├─ references/deep detail pulled in only when needed — progressive disclosure
├─ scripts/real code, so the agent composes instead of rebuilding boilerplate
├─ assets/templates & files to copy into the output
├─ config.jsonsetup the agent asks for if it’s missing (e.g. which Slack channel)
└─ hooks + memoryon-demand guardrails + an append-only log so it remembers
Why it matters: the folder itself is the knowledge base. The agent reads the root, then reaches deeper only when the task demands it — the same way you’d hand a new hire a one-pager that points to the detailed docs.
The nine types — a gap-analysis map for your own library
1Library / API reference
2Product verification ★ top impact
3Data fetching & analysis
4Business-process automation
5Code scaffolding & templates
6Code quality & review
7CI/CD & deployment
8Runbooks
9Infrastructure operations
By Anthropic’s own measurement, verification Skills — the ones that check the work — moved output quality the most. If you build one category well, build that one.
The craft — what separates a good Skill from a useless one
Gotchas = highest-signal section Describe for the model, not humans (it’s the trigger) Don’t state the obvious Ship scripts, not just prose On-demand guardrail hooks (/careful, /freeze) Let it remember (log / SQLite) Don’t railroad — leave room to adapt
The take

The knowledge of how your organization actually operates can be captured, versioned, shared & executed — and the thing capturing it is a humble folder with a script and a gotchas list inside. For the builder, that’s context engineering with real tools attached. For whoever owns the budget, it’s the difference between AI that starts from zero every morning and an asset that compounds. Caveats: best practices are still evolving, checked-in Skills cost context, and curation beats accumulation. Start with one Skill, one gotcha, and the category that catches your mistakes.

Source: “Lessons from building Claude Code: How we use skills,” Thariq Shihipar (Anthropic), Claude blog, 3 June 2026. Categories, examples & measured claims are Anthropic’s; framing is the author’s. Docs: code.claude.com/docs/en/skills.
thorstenmeyerai.com

Implications for AI Development and Organizational Knowledge

This development signals a shift toward more durable, scalable, and maintainable AI systems within organizations. By packaging knowledge into reusable folders, companies can achieve greater consistency in output, accelerate onboarding, and build an evolving library of institutional memory. For enterprises, this approach reduces reliance on ad-hoc prompts and makes AI capabilities more aligned with real-world workflows, ultimately improving operational reliability and efficiency.

Moreover, the emphasis on verification Skills and structured workflows addresses common AI quality issues, such as mistakes and inconsistencies, which are critical for deploying AI in high-stakes environments. The approach encourages organizations to view Skills as assets that appreciate over time, rather than disposable prompts, fostering a more strategic investment in AI infrastructure.

Create a Podcast with AI (No Experience Needed) : A Step-by-Step Guide to Planning, Scripting, Recording, Editing, and Launching a Podcast Using ChatGPT, AI Tools, and Automation

Create a Podcast with AI (No Experience Needed) : A Step-by-Step Guide to Planning, Scripting, Recording, Editing, and Launching a Podcast Using ChatGPT, AI Tools, and Automation

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Background of AI Skill Development and Organizational Practices

Most organizations using AI coding agents have relied on manually crafting prompts for each task, leading to inconsistency and onboarding challenges. Anthropic’s recent internal experiments with hundreds of Skills—defined as comprehensive folders—represent a departure from this norm. This approach aligns with broader industry efforts to embed organizational knowledge directly into AI systems, moving beyond ephemeral prompts toward durable, reusable assets.

Prior to this, the common practice was to treat prompts as simple instructions, often retyped daily or reused with minor variations. Anthropic’s research demonstrates that structuring knowledge as folders containing scripts, reference materials, and configurations significantly enhances the robustness and scalability of AI workflows, especially in complex enterprise environments.

This shift reflects a broader trend toward institutionalizing AI capabilities, making them part of everyday business operations rather than ad-hoc solutions. It also builds on previous insights about prompt engineering, emphasizing the importance of specific, non-obvious instructions and error handling to improve output quality.

“A Skill is a folder — one that can contain instructions, reference documents, runnable scripts, templates, data, configuration, and even hooks that fire only while the Skill is active.”

— Thorsten Meyer, AI researcher at Anthropic

AI for Bookkeeping Automation and Workflows: Automate Data Entry, Receipts, Categorization, Reconciliation, and Month-End Reporting Using AI and No-Code Tools, Save Hours Every Week for Bookkeepers

AI for Bookkeeping Automation and Workflows: Automate Data Entry, Receipts, Categorization, Reconciliation, and Month-End Reporting Using AI and No-Code Tools, Save Hours Every Week for Bookkeepers

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Unanswered Questions About Skill Implementation and Scaling

It is not yet clear how widely other organizations will adopt the folder-based Skill approach or how it performs outside Anthropic’s internal environment. Details about the specific technical implementations, such as how agents discover and activate Skills in complex workflows, remain proprietary or under development. Additionally, the long-term scalability and maintenance of Skills libraries as they grow remain to be seen.

The AI Transformation Starter Guide - Everything You Need To Know To Get Started With AI In Your Business And Daily Life: Master Artificial ... ... Generative AI for Business and Creativity)

The AI Transformation Starter Guide – Everything You Need To Know To Get Started With AI In Your Business And Daily Life: Master Artificial … … Generative AI for Business and Creativity)

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Future Steps for Broader Adoption and Technical Refinement

Anthropic is likely to continue refining its Skills framework and share more detailed methodologies for external adoption. Other organizations may begin experimenting with similar folder-based structures, and industry standards could evolve around this approach. Monitoring how Skills are integrated into larger operational systems and how they impact AI reliability will be key in the coming months.

Furthermore, future updates may include tools for easier Skills management, versioning, and automation, as well as broader case studies demonstrating effectiveness in varied enterprise scenarios.

Google’s Agent Development Kit Explained: Building, Securing, and Scaling AI Agents with Google ADK: From Core Concepts and Deployment to Advanced Architectures and Domain Applications

Google’s Agent Development Kit Explained: Building, Securing, and Scaling AI Agents with Google ADK: From Core Concepts and Deployment to Advanced Architectures and Domain Applications

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

How does a Skill differ from a prompt in AI systems?

A Skill is a folder containing instructions, scripts, and assets that define a durable, reusable workflow, whereas a prompt is a simple instruction or question used temporarily to guide the AI.

Why are verification Skills considered most valuable?

Verification Skills check the AI’s output for mistakes, significantly improving output quality and reducing errors, especially in critical applications.

Can other organizations implement this folder-based Skill approach?

While Anthropic’s experience is promising, it remains to be seen how easily this method can be adopted broadly. Technical details and organizational readiness will influence adoption success.

What are the main benefits of using Skills as folders?

Skills as folders improve consistency, accelerate onboarding, and allow continuous improvement by capturing organizational knowledge in a structured, reusable format.

What challenges might organizations face in adopting Skills as folders?

Potential challenges include integrating this approach into existing workflows, managing version control, and developing the technical infrastructure for discovery and execution.

Source: ThorstenMeyerAI.com

You May Also Like

The Local-First Agentic Operator

A single operator using agentic AI now builds and manages multiple complex products across domains, traditionally requiring organizations, marking a shift in software development.

ChannelHelm: One Video, Every Platform

ChannelHelm automates the creation of multi-platform content from a single video, reducing manual effort and expanding reach efficiently.

The Compounding Error Problem — Why 99.9% Alignment Decays to 60% in 500 Generations

Analysis of how 99.9% alignment accuracy per generation degrades to 60% after 500 generations, highlighting risks for recursive AI self-improvement.

CarPlay Is Additive

New insights reveal that CarPlay enhances vehicle functionality without replacing existing systems, raising questions about its long-term impact.