5 Skill Hacks Nobody Talks About 🛠️

Before you build anything from scratch, know where to find skills that already exist. Most people skip this and waste time rebuilding what's already been done.

• Anthropic's built-in library: Go to Settings → Customize → Skills → Browse. There's an entire library of pre-made skills you can install in one click. Most people don't even know this exists.
• Community plugins come with skills: When you install a Plugin (Settings → Customize → Plugins), it often bundles 3-5 related skills. Install a plugin, get a skill pack for free.
• 10 ready-to-use work skills: I built 10 professional skills you can copy and paste right now — status reports, meeting recaps, email drafts, and more.
• Turn any good Cowork chat into a skill: Already gave Claude good instructions in a Cowork session? Click the arrow next to the chat name → "Turn into a skill." Done. No building required.

Skill not firing when it should? Don't guess. Ask Claude:

Claude reads the skill description back to you exactly as written. You'll instantly see why it's not matching. 90% of skill problems are in the description, not the instructions. Fix what Claude reads back, and the skill starts working.

Most broken skills aren't broken — they're firing at the wrong time. The fix is one line in the description:

What you tell a skill NOT to do matters more than what you tell it to do. Without this line, Claude will try to use your "email draft" skill every time someone mentions email in any context. Add this to every single skill you build.

This is the most underused feature in Claude. You don't need to build skills from scratch.

Open any Cowork chat where you already gave Claude good instructions → click the arrow next to the chat name → click "Turn into a skill."

Claude packages everything into a reusable skill automatically. Every good Cowork conversation you've ever had is a skill waiting to be saved. Go through your Cowork history right now — you probably have 5-10 skills already sitting there.

People hold back on installing skills because they think it'll slow Claude down or cause conflicts. It won't.

Claude only reads the full instructions when a skill matches your current task. Everything else is completely ignored. You can have 300 skills installed and barely use any extra tokens. The cost of an unused skill is literally zero.

Install every skill that's even remotely relevant. Browse the library, install plugins, save your Cowork chats. Let Claude sort out which one to use — it's better at matching than you think.

If Claude starts cutting corners, rushing through steps, or giving you half-effort output inside a skill — don't change the skill file. Add this to your actual message:

Anthropic confirmed this works better in the prompt than in the skill file itself. The skill sets what to do. Your message sets how hard to try. Two different levers — use both.

• Test every skill immediately after building it. Don't assume it works. Send it a real task right away and check the output.
• Name skills with slash commands. Use names like /inbox, /prep, /recap, /draft. You can type the slash command to trigger the skill instantly instead of waiting for Claude to auto-detect it.
• Keep instructions specific. "Write a 3-paragraph email in a professional but warm tone" beats "write an email." The more specific your skill instructions, the less you have to correct in the output.
• Update skills monthly. Your workflows change. Your priorities shift. Set a monthly reminder to review your top 5 skills and update anything that's stale.
• Use skills inside Projects. A skill running inside a Project gets all the Project's context for free — your files, your instructions, your preferences. Same skill, dramatically better output.
• Chain skills together. Run /inbox-triage first, then /meeting-prep for anything flagged as urgent. Skills don't have to work alone — they're building blocks.
• Skill not working? Read Why Your Claude Skill Isn't Working — it covers the 4 most common reasons and how to fix each one.