Skip to main content
The AI Breakdown

How to Use Agent Skills

27 min episode · 2 min read

Episode

27 min

Read time

2 min

Topics

Fundraising & VC, Artificial Intelligence, Software Development

AI-Generated Summary

Key Takeaways

  • Progressive Disclosure Architecture: Skills use a three-layer loading system — a ~100-token metadata description, the full skill.md body, and linked supplementary files — so agents load only the context needed at each decision point. This prevents the system prompt bloat that caused earlier agents to become slower, more expensive, and less reliable as capabilities expanded.
  • The Gotcha Section: The highest-signal content in any skill is a dedicated section documenting common failure points Claude hits when executing that skill. Update this section each time the agent makes a mistake, turning the skill into a living document that accumulates institutional knowledge and prevents repeated errors over time.
  • Two Skill Categories for Testing Strategy: Skills fall into capability uplift (Claude can't do this reliably without the skill) or encoded preference (Claude can do each step, but the skill sequences them to match team workflows). Capability uplift skills may become obsolete as models improve; encoded preference skills are more durable but only as valuable as their fidelity to actual workflows.
  • Skill Creator Tool for Non-Engineers: Anthropic updated their skill creator to let subject matter experts — not just engineers — test and benchmark skills without writing code. It runs evals against multiple prompts, scores performance, runs A/B tests against base Claude, and auto-rewrites vague descriptions. Anthropic tested this on their own skills and saw improved triggering in five out of six cases.
  • Skills as Cross-Platform Reusable Capabilities: Skills are supported across Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and now Notion AI, meaning a skill authored once works across ecosystems. Notion's implementation lets users convert any page into a skill with one click, signaling that the reusable-capability model is converging across the entire AI stack from consumer to enterprise.

What It Covers

The Claude Code team at Anthropic shares how they build and use agent skills — reusable folders of instructions, scripts, and resources that load contextually rather than bloating system prompts. The episode covers skill architecture, nine key skill categories, best practices from Tariq's post, and how the concept applies across all user levels.

Key Questions Answered

  • Progressive Disclosure Architecture: Skills use a three-layer loading system — a ~100-token metadata description, the full skill.md body, and linked supplementary files — so agents load only the context needed at each decision point. This prevents the system prompt bloat that caused earlier agents to become slower, more expensive, and less reliable as capabilities expanded.
  • The Gotcha Section: The highest-signal content in any skill is a dedicated section documenting common failure points Claude hits when executing that skill. Update this section each time the agent makes a mistake, turning the skill into a living document that accumulates institutional knowledge and prevents repeated errors over time.
  • Two Skill Categories for Testing Strategy: Skills fall into capability uplift (Claude can't do this reliably without the skill) or encoded preference (Claude can do each step, but the skill sequences them to match team workflows). Capability uplift skills may become obsolete as models improve; encoded preference skills are more durable but only as valuable as their fidelity to actual workflows.
  • Skill Creator Tool for Non-Engineers: Anthropic updated their skill creator to let subject matter experts — not just engineers — test and benchmark skills without writing code. It runs evals against multiple prompts, scores performance, runs A/B tests against base Claude, and auto-rewrites vague descriptions. Anthropic tested this on their own skills and saw improved triggering in five out of six cases.
  • Skills as Cross-Platform Reusable Capabilities: Skills are supported across Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and now Notion AI, meaning a skill authored once works across ecosystems. Notion's implementation lets users convert any page into a skill with one click, signaling that the reusable-capability model is converging across the entire AI stack from consumer to enterprise.

Notable Moment

Anthropic found that despite roughly 28,000 skills existing on ClawHub, the vast majority fit into just nine categories — a surprisingly narrow taxonomy given the volume. The team concluded that most agent work clusters around a predictable set of recurring task types regardless of organization or industry.

Know someone who'd find this useful?

Episode Transcript

Today on the AI Daily Brief, how the team that designed agent skills uses agent skills, and before that in the headlines, you can now control Claude Cowork from your phone. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. Alright, friends. Quick announcements before we dive in. First of all, thank you to today's sponsors, KPMG, Blitsy, AIUC, and Mercury. To get an ad free version of the show, go to patreon.com/aidailybrief, or you can subscribe on Apple Podcasts. Ad free is just $3 a month. If you are interested in sponsoring the show, send us a note at sponsors@aidailybrief.ai. At this point, we are firmly selling into the summer, so if you are planning campaigns in the future, it is a good time to reach out. And of course, if you need to know anything else about the ecosystem, you can also find that on a I daily brief dot a I. I would once again point you to the newsletter which is back and is basically the best way to get access to the links that I talk about in the show. Again, you can get that all on a idailybrief.ai, and with that out of the way, let's dive in. One of the interesting ways that you can tell what's really important to AI builders and people on the front lines is when there's a story that on the surface looks fairly small, but which is getting a disproportionate share of the conversation in AI circles. Our fourth story today is exactly that. On the surface, it's just a simple new feature for Claude CoWork. In this case, it's called dispatch, and it allows you to bring your Claude CoWork with you on the go. That said, based on the reaction, 3,000,000 views on the announcement tweet, 9,000 bookmarks, this one is a big deal to people. In the wake of OpenClaw, companies in the agent space have either been, a, releasing their own versions of OpenClaw, that was obviously the topic of our show yesterday, or they've been slowly adding the important features of OpenClaw to their existing product suites, which has been, of course, Anthropic's approach. A couple weeks ago, we got remote control for Claude Code, which allowed users to initiate Claude sessions on their computer and then carry them onto their mobile devices where they could control them doing whatever it was that they were doing, basically coding from the gym. Dispatch is basically that but for co work. The co work sessions are still hosted in the sandbox on your computer, meaning Claude still has the same access and protections. However, you can now kick off a co work session and then continue monitoring progress and providing approvals while out and about. Anthropic described the feature as like having a walkie talkie for communicating with Claude. Co work developer, Felix Reisberg wrote, it feels pretty magical to give Claude a mission …

Get the full transcript (5,643 words) + summary by email — free

One-time email with the complete transcript and AI summary of this episode. No account needed.

One email, no spam. We’ll also show you what SignalCast does.

Browse all The AI Breakdown transcripts →

You just read a 3-minute summary of a 24-minute episode.

Get The AI Breakdown summarized like this every Monday — plus up to 2 more podcasts, free.

Pick Your Podcasts — Free

Keep Reading

Books, tools, and gear mentioned in this episode

SignalCast may earn commission on purchases via these links.

Tools

  • by Anthropic

    Anthropic found that despite roughly 28,000 skills existing on ClawHub, the vast majority fit into just nine categories — a surprisingly narrow taxonomy given the volume.
  • by GitHub

    Skills are supported across Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and now Notion AI, meaning a skill authored once works across ecosystems.
  • by Notion

    Skills are supported across Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and now Notion AI, meaning a skill authored once works across ecosystems. Notion's implementation lets users convert any page into a skill with one click.
  • SPONSORS: Blitzy
  • SPONSORS: Mercury
  • by Anthropic

    The Claude Code team at Anthropic shares how they build and use agent skills — reusable folders of instructions, scripts, and resources that load contextually rather than bloating system prompts.
  • by OpenAI

    Skills are supported across Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and now Notion AI, meaning a skill authored once works across ecosystems.
  • Skills are supported across Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and now Notion AI, meaning a skill authored once works across ecosystems.

company

  • SPONSORS: KPMG
  • SPONSORS: AIUC

More from The AI Breakdown

We summarize every new episode. Want them in your inbox?

Similar Episodes

Related episodes from other podcasts

Explore Related Topics

This podcast is featured in Best AI Podcasts (2026) — ranked and reviewed with AI summaries.

Read this week's AI & Machine Learning Podcast Insights — cross-podcast analysis updated weekly.

You're clearly into The AI Breakdown.

Every Monday, we deliver AI summaries of the latest episodes from The AI Breakdown and 192+ other podcasts. Free for one show.

Start My Monday Digest

No credit card · Unsubscribe anytime