A Skill Is A Folder, Not A Prompt: What Anthropic Learned Running Hundreds Of Them
AIThis post was created with the assistance of artificial intelligence (AI).

📊 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.

Before you orderOffer from Amazon

Get smart everyday buys delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

Anthropic has shifted from viewing AI Skills as prompts to treating them as folders containing instructions, scripts, and assets. This approach enhances consistency, onboarding, and institutional knowledge. The company ran hundreds of these Skills internally, emphasizing their value as evolving organizational assets.

Anthropic has announced a significant shift in how it develops and deploys AI Skills, now treating them as folders containing instructions, scripts, and reference assets rather than just prompts. This approach aims to improve consistency, onboarding, and institutional knowledge, marking a departure from common prompt-based methods used across AI teams.

In a detailed write-up from a Claude Code engineer, Anthropic explained that a Skill is fundamentally a container—a folder that can hold instructions, reference documents, runnable scripts, templates, configuration data, and hooks. This structure allows AI agents to discover, read, and execute the contents inside, creating a more durable and reusable organizational asset.

Anthropic’s internal experiments involved running hundreds of such Skills across its engineering teams, with the company emphasizing that this method transforms ad-hoc prompting into a standardized, version-controlled process. The approach facilitates consistent output, simplifies onboarding of new team members, and enables continuous improvement of Skills through iteration.

The company identified nine core categories of Skills, from library references and product verification to infrastructure operations, with verification Skills deemed most valuable for quality control. The emphasis is on building Skills that catch mistakes and push models off default behaviors, rather than restating obvious instructions.

Technical lessons include avoiding redundant prompts, focusing on non-obvious knowledge, and crafting precise description triggers that match user requests accurately. Bundling real code and helper functions within Skills enhances their utility and reusability, making them a core part of organizational workflows.

At a glance
reportWhen: published March 2024
The developmentAnthropic published insights from its internal use of Skills, demonstrating a new approach where Skills are structured as folders rather than simple prompts, improving AI reliability and organizational knowledge sharing.
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 Business Operations

This development signifies a shift towards more durable, scalable, and reliable AI deployment practices within organizations. By structuring Skills as folders, companies can better standardize processes, reduce onboarding time, and create a living library of institutional knowledge. This approach also positions Skills as strategic assets that improve over time, rather than static prompts that decay in usefulness.

For businesses, adopting this method could lead to more consistent AI outputs across teams, better error handling, and more efficient knowledge transfer. It also encourages a disciplined, versioned approach to AI asset management, potentially transforming operational workflows and reducing reliance on ad-hoc prompt engineering.

Products Worth Considering

Logitech MX Master 4 Ergonomic Wireless Mouse with Haptics – Graphite

Logitech MX Master 4 Ergonomic Wireless Mouse with Haptics – Graphite

Apple MacBook Air 15-inch (2026) with M5 chip, 16GB memory, 512GB SSD — Midnight

Apple MacBook Air 15-inch (2026) with M5 chip, 16GB memory, 512GB SSD — Midnight

As an affiliate, we earn on qualifying purchases.

From Prompting to Asset Management in AI Teams

Historically, AI teams have relied heavily on prompt engineering—crafting specific instructions for each task. While effective at times, this method often leads to inconsistent results, onboarding difficulties, and a lack of institutional memory. Anthropic’s new approach, shared by a Claude Code engineer, builds on lessons learned from running hundreds of Skills internally, emphasizing a shift towards structured, reusable organizational assets.

This move aligns with broader trends in AI deployment, where companies seek more scalable and maintainable methods for integrating AI into workflows. The concept of Skills as folders reflects a maturation of AI asset management, akin to software libraries or operational playbooks.

Prior to this, most organizations lacked a formalized system for capturing and sharing the nuanced knowledge that guides AI behavior, often relying on brittle prompt chains or informal documentation. Anthropic’s approach offers a way to institutionalize this knowledge, making it more durable and accessible.

“Treating Skills as folders containing instructions and scripts fundamentally changes how organizations can develop and maintain AI capabilities.”

— Thorsten Meyer, AI researcher

Unclear Aspects of the Folder-Based Skill System

It is not yet clear how widely Anthropic plans to implement this approach outside its internal teams or how other organizations will adapt the folder-based Skill model. The long-term effectiveness and scalability of this method remain to be seen, especially in different operational contexts or with less technical teams.

Additionally, details about how Skills are versioned, shared across teams, and integrated into existing workflows are still emerging. The specific technical standards and tooling support required for broader adoption are also not fully known.

Next Steps for Adoption and Industry Impact

Anthropic is expected to continue refining its Skills framework and may publish more detailed guidelines or tooling to facilitate adoption. Other AI practitioners and organizations will likely observe these developments closely, experimenting with similar structures to improve their own AI deployment practices.

Further research and case studies will be needed to evaluate how this approach impacts operational efficiency, model quality, and organizational knowledge management over time.

Key Questions

How does treating Skills as folders improve AI consistency?

By bundling instructions, scripts, and reference assets in a structured folder, Skills provide a durable, version-controlled way to ensure AI outputs are consistent across different runs and teams.

What are the main categories of Skills identified by Anthropic?

They include library references, product verification, data analysis, business automation, code scaffolding, quality review, CI/CD, runbooks, and infrastructure operations.

Can this approach be adopted by smaller organizations?

While technically feasible, smaller teams may need to adapt the complexity of the folder structure and tooling to fit their resources and workflows.

What technical lessons did Anthropic learn about building effective Skills?

Key lessons include avoiding redundant prompts, focusing on non-obvious knowledge, crafting precise trigger descriptions, and bundling real code for reusability.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

Speech By Commissioner Hoekstra At The Launch Of Europe’s Largest Carbon Capture And Storage Project At Yara Sluiskil, The Netherlands

EU Commissioner Hoekstra announced the launch of Europe’s largest carbon capture and storage initiative at Yara Sluiskil, Netherlands, highlighting Europe’s climate efforts.

Meta’s AI Coding Strategy Gets A Boost With Muse Spark 1.2

Meta releases Muse Spark 1.2 and Muse Code, its first co-trained coding model and agent, aiming to compete with OpenAI and Anthropic in AI coding tools.

Apple Wants Blacklisted Chinese RAM — and That Tells You How Bad the Squeeze Got

Apple is lobbying US authorities to buy Chinese-made memory chips from CXMT, raising questions about supply chain and national security amid a global memory crunch.

Summer Trend Alert: Portland’s Daylight Reaches Nearly 15 Hours

Portland experiences nearly 15 hours of daylight during the summer solstice, a confirmed astronomical event affecting local routines and planning.