TL;DR
Anthropic has published lessons from using hundreds of Claude Code Skills across its engineering organization. The company frames Skills as reusable folders containing instructions, scripts, references and guardrails, not just saved prompts.
Anthropic has published lessons from using hundreds of Claude Code Skills inside its engineering organization, saying the format helped teams package repeated agent instructions into shared, versioned workflows rather than rewriting prompts for each task.
The post, “Lessons from building Claude Code: How we use skills,” was written by Thariq Shihipar and published on Anthropic’s Claude blog on June 3, 2026, according to the source material. A July 1 Thorsten Meyer AI dispatch framed the update as evidence that AI agent work is moving from informal prompting toward reusable operational assets.
Anthropic describes a Skill as a folder the agent can discover and use. That folder can include SKILL.md instructions, reference files, scripts, templates, configuration and hooks. The source says the agent reads the root instructions first, then pulls in more material only when the task requires it.
Anthropic’s internal Skills reportedly clustered into nine categories: library and API references, product verification, data fetching and analysis, business-process automation, code scaffolding and templates, code quality and review, CI/CD and deployment, runbooks, and infrastructure operations. The company’s own measurement, as summarized by the source, found verification Skills had the strongest effect on output quality.
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.
“A Skill is just a clever markdown prompt you save in a file.”
A folder the agent can discover, read & run — instructions, scripts, references, templates, config & on-demand hooks.
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.
Reusable Agent Workflows Gain Weight
The development matters because it shows how companies using AI coding agents may shift from individual prompt craft to institutional workflow design. If a Skill captures how a company reviews code, verifies product behavior or handles deployment steps, that knowledge can be shared and updated like other engineering assets.
For engineering teams, the value is consistency: the same agent task can draw on the same scripts, templates and caveats regardless of who runs it. For managers, the claim is that a Skills library may become a durable record of how work is done, rather than a collection of one-off instructions scattered across chats or personal notes.
Agentic AI Engineering: Building AI Agents for Beginners: A Hands-On Guide to No-Code Workflows, LLM Tools, RAG, Automation, and Safe Multi-Agent Systems
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Claude Code’s Folder-Based Pattern
The source material says the main correction in Anthropic’s post is definitional: a Skill is not just markdown. It is a file-system unit with instructions and supporting assets that an agent can read and, when needed, execute through included scripts.
That design uses what the dispatch calls progressive disclosure: short root instructions guide the model, while deeper references are loaded only for relevant tasks. In practice, that means a Skill can act like a compact operating guide pointing to longer documentation, reusable code and task-specific guardrails.
“A Skill is not a clever prompt saved in a text file. It’s a folder.”
— Thorsten Meyer AI dispatch
![MixPad Free Multitrack Recording Studio and Music Mixing Software [Download]](https://m.media-amazon.com/images/I/71ltIxIuz1L._SL500_.jpg)
MixPad Free Multitrack Recording Studio and Music Mixing Software [Download]
Create a mix using audio, music and voice tracks and recordings.
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Open Questions For Adoption
Several details remain unclear from the source material. Anthropic’s exact measurement method for judging output quality is not described in full, and the source does not provide the underlying data behind the claim that verification Skills had the largest effect.
It is also unclear how easily Anthropic’s results transfer to smaller teams, non-engineering groups or companies with less mature internal documentation. The source notes that best practices are still evolving, and that checked-in Skills can add context cost if teams accumulate them without curation.
versioned AI instruction folders
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Teams Test Skills Libraries
The next step for teams using Claude Code is likely experimentation with a small number of high-value Skills, especially those that check work, encode recurring review steps or package repeated setup tasks. The source recommends starting with one Skill, one hard-won caveat and a category that catches common mistakes.
Anthropic’s public documentation at code.claude.com/docs/en/skills is the reference point for teams that want to build Skills using the company’s current format. More evidence will be needed to show how much these libraries improve quality, speed and onboarding outside Anthropic’s own engineering environment.
Agentic Development: The Complete Guide to AI-Assisted Coding with Claude, Cursor, and Beyond
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did Anthropic announce about Skills?
Anthropic published a June 3, 2026 Claude blog post describing lessons from using hundreds of Claude Code Skills across its engineering organization.
What is a Claude Code Skill?
A Skill is described as a folder that can contain instructions, references, scripts, templates, configuration and hooks for an AI coding agent to use during a task.
Why are verification Skills important?
According to Anthropic’s reported measurement, verification Skills had the largest effect on output quality because they help agents check their work rather than only generate it.
Is this only useful for developers?
The current example comes from Claude Code and Anthropic’s engineering organization, but the broader idea is reusable workflow knowledge. How well it works outside software teams remains not fully proven from the source material.
What should teams try first?
The source recommends beginning with one focused Skill, especially in a category where mistakes are common, and improving it as new edge cases appear.
Source: Thorsten Meyer AI
