Anthropic's Claude Cowork Lets You Teach AI by Recording Your Screen
Claude Cowork's new 'Record a skill' feature lets you demonstrate a workflow once on screen, and Claude turns it into a reusable automation it can run on demand.
- New feature: Claude Cowork's Record a skill lets you record your screen while narrating a task, and Claude converts it into a reusable automation skill.
- Where to find it: Available in the
+menu of the Claude desktop app under "Record a skill", on Pro, Max, and Team plans. - What skills are: Structured instruction packs (a folder with a
SKILL.mdfile) that Claude loads automatically when a relevant task comes up. - Who it's for: Primarily non-technical knowledge workers; Anthropic data shows 90%+ of Cowork usage is non-software work (ops, finance, content).
- Key limitation: Cowork tasks consume significantly more usage allocation than regular chat, and Claude can modify files without prior review if not configured carefully.
- Team sharing: On Team and Enterprise plans, recorded skills can be shared with specific colleagues or published to the entire organization's directory.
Writing instructions for an AI agent is tedious. You have to describe every step, anticipate edge cases, and hope the model interprets your words the way you meant them. Anthropic just shipped a shortcut: record your screen while you do the task, talk through what you're doing, and Claude Cowork converts that into a reusable skill it can run again on demand.
Record once, reuse forever
The new Record a skill feature lives in the + menu of the Claude desktop app. Hit record, do the task on your screen while narrating your reasoning out loud, stop the recording, and Claude synthesizes everything into a skill it can replay later. No writing required, no markdown files to author by hand.
The previous path to creating a custom skill was more technical. Skills are structured instruction packs that Claude loads when relevant, and until now, creating one meant writing a SKILL.md file yourself: a structured markdown document with metadata, step-by-step instructions, and optional executable scripts. That approach gives you control and transparency, but it requires writing, and non-technical users may find the structure intimidating. The recording flow inverts that: the skill is authored for you from a demonstration.