A Skill Is A Folder, Not A Prompt: What Anthropic Learned Running Hundreds Of Them

📊 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 shared a new approach to building AI skills, emphasizing that skills are best understood as folders containing instructions, scripts, and assets rather than simple prompts. This method improves consistency, onboarding, and asset development, marking a shift in enterprise AI practices.

Anthropic has revealed that its approach to building AI skills involves structuring them as folders containing instructions, scripts, and assets, rather than simple prompts. This shift aims to improve the consistency, reusability, and institutional memory of AI capabilities within organizations, marking a significant departure from traditional prompt engineering methods.

According to a detailed write-up from an Anthropic Claude Code engineer, a Skill is defined as a folder that can include instructions, reference documents, runnable scripts, templates, data, configuration, and hooks. This structure allows AI agents to discover, read, and execute the contents dynamically, making skills more durable and context-aware.

Anthropic emphasizes that this approach transforms ad-hoc prompting into a standardized, institutional capability. Skills serve as reusable assets that encapsulate tribal knowledge, guardrails, and operational procedures, rather than mere sticky notes or prompts. This methodology enables organizations to ensure consistent outputs, streamline onboarding, and develop a growing library of refined skills.

Anthropic’s internal analysis identified nine categories of Skills, including library references, product verification, data analysis, automation, code scaffolding, review, deployment, runbooks, and infrastructure operations. The company highlights that verification Skills, which check the work output, are among the most valuable, as they directly impact quality and error reduction.

The technical core of building effective Skills involves avoiding redundancy, focusing on non-obvious, specific content, and writing precise trigger descriptions that match real user requests. Bundling real code and helper functions within Skills enhances their utility and reduces the need for manual intervention.

At a glance
reportWhen: published recently, based on Anthropic’…
The developmentAnthropic published a detailed internal report explaining that their AI skills are structured as folders with instructions, scripts, and assets, rather than just prompts, influencing how organizations develop AI capabilities.
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 Enterprise AI Development

This development signals a shift in how organizations can manage and scale AI capabilities. By treating Skills as structured folders rather than prompts, companies can create more reliable, maintainable, and sharable AI assets. This approach enhances consistency across teams, accelerates onboarding, and fosters continuous improvement through iterative refinement of Skills.

It underscores the importance of institutional memory in AI operations, reducing reliance on individual knowledge and promoting best practices. As Skills become assets that grow more sophisticated over time, organizations can better leverage AI for complex, operational tasks, and improve overall output quality.

Furthermore, this perspective elevates the role of technical craftsmanship in AI development, emphasizing precise documentation, specific instructions, and code bundling. It encourages a more disciplined, asset-driven approach rather than one based on ephemeral prompts, which can be fragile and inconsistent.

Amazon

AI development automation scripts

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

From Prompt Engineering to Asset-Based AI Building

Traditionally, AI development has focused on prompt engineering—crafting specific instructions to guide model outputs. However, this method often results in fragile interactions that require constant tweaking. Anthropic’s internal documentation suggests a paradigm shift towards treating AI skills as structured assets, akin to software modules or operational procedures.

This approach builds on prior efforts to systematize AI capabilities but emphasizes reusability, versioning, and institutional knowledge. The concept aligns with broader trends in enterprise AI, where organizations seek scalable, maintainable solutions that can adapt to evolving needs.

Anthropic’s insights come after extensive internal experiments with hundreds of Skills, revealing that categorizing and refining them into a library improves performance and reduces errors. The nine-category framework provides a practical roadmap for organizations to identify gaps and develop their own Skills libraries.

“Treating Skills as folders containing scripts and assets rather than prompts fundamentally changes how organizations build, share, and maintain AI capabilities.”

— Thorsten Meyer, AI researcher

Amazon

enterprise AI skills management tools

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Unresolved Questions About Skill Implementation

It remains unclear how widely adopted this folder-based Skill model will be across different organizations and industries. Details about the implementation costs, integration challenges, and long-term maintenance are still emerging. Additionally, the precise tooling and processes needed to manage large Skills libraries at scale are not yet fully documented.

Further research is needed to assess how this approach compares in practice to traditional prompt engineering in terms of efficiency, cost, and output quality over time.

Amazon

AI instruction and asset organization software

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Next Steps for Organizations Adopting the Folder-Based Model

Organizations interested in this approach should start by cataloging their current Skills, identifying gaps based on the nine categories, and experimenting with folder-based structures. Developing internal tools for managing, versioning, and sharing Skills will be crucial. Watching how Anthropic refines its own library and shares best practices will inform broader adoption.

Further developments may include standardized frameworks, tooling support, and case studies demonstrating the economic and operational benefits of this model, shaping future enterprise AI strategies.

Amazon

AI workflow automation templates

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

What is a Skill in Anthropic’s new approach?

A Skill is a structured folder containing instructions, scripts, reference documents, templates, data, and hooks, designed to be discovered and executed by AI agents, rather than just a prompt or note.

How does this change the way organizations build AI capabilities?

It shifts the focus from crafting ephemeral prompts to developing reusable, versioned assets that encapsulate tribal knowledge, guardrails, and operational procedures, leading to more consistent and maintainable AI systems.

What are the main categories of Skills identified by Anthropic?

The nine categories include library references, product verification, data analysis, automation, code scaffolding, review, deployment, runbooks, and infrastructure operations.

What challenges might organizations face adopting this model?

Potential challenges include developing tooling for managing large Skills libraries, integrating with existing workflows, and ensuring continuous updates and refinement of assets over time.

Why is the verification Skills category considered most valuable?

Because verification Skills directly impact output quality by catching errors and mistakes, thus reducing costly errors and improving reliability.

Source: ThorstenMeyerAI.com

You May Also Like

ShinyHunters · The New APT Model.

ShinyHunters has evolved into a distributed, AI-enabled threat collective operating as a scalable Extortion-as-a-Service model, surpassing traditional APTs.

VigilSAR Benchmark: There Is No Best Model

VigilSAR Benchmark reveals there is no universally best AI model for defense, emphasizing context-specific rankings based on capability, reliability, and compliance.

Fair-value appraisals for used GPUs and AI hardware

A new manual valuation tool aims to establish fair market prices for used data-center GPUs and AI hardware, addressing pricing disputes in the secondary market.

Anchor. The Schwarz Group model.

An in-depth analysis of Schwarz Group’s €11B AI data center investment and its potential as a scalable European industrial-anchor model.