{"id":40323,"date":"2026-01-29T06:42:15","date_gmt":"2026-01-29T06:42:15","guid":{"rendered":"https:\/\/adspyder.io\/blog\/?p=40323"},"modified":"2026-01-29T08:49:53","modified_gmt":"2026-01-29T08:49:53","slug":"how-to-add-ai-skills","status":"publish","type":"post","link":"https:\/\/adspyder.io\/blog\/how-to-add-ai-skills\/","title":{"rendered":"How to Add a Skill to Claude and Alexa (Plus skills.sh for Agents) in 2026"},"content":{"rendered":"

<\/p>\n

\n
\n

<\/p>\n

The most searched \u201cskills\u201d questions aren\u2019t philosophical\u2014they\u2019re practical: add skill to Claude<\/span><\/strong>, adding a skill to Alexa<\/span><\/strong>, and the broader curiosity around Gemini skills<\/span> and ChatGPT skills<\/span>. Translation: marketers want assistants that can run repeatable workflows<\/b>, not generate one-off answers. This guide shows you how to add AI skills to Claude across ecosystems (they\u2019re not just a Claude thing), and gives you clean, step-by-step instructions for\u00a0adding a skill to Claude Code<\/b>, installing skills using the open ecosystem at skills.sh<\/a><\/b>, and building\/publishing an Alexa skill via the Alexa Skills Kit<\/b>.
\nYou\u2019ll also get a marketer-friendly checklist and a simple \u201cbuild vs buy\u201d decision framework.<\/p>\n

<\/p>\n

\n
\n
Want to turn competitor research into a reusable \u201cskill\u201d?<\/div>\n
Pull real creatives and landing pages in AdSpyder, then standardize your analysis into a repeatable workflow your team can run every week.<\/div>\n<\/div>\n

Explore AdSpyder \u2192<\/a><\/p>\n<\/div>\n

<\/p>\n

\n
\n
\n
Table of contents<\/div>\n<\/div>\n
Jump to a section<\/div>\n<\/div>\n
What \u201cskills\u201d mean<\/a>
\nKey statistics<\/a>
\nDecision flow<\/a>
\nAdd a skill to Claude Code<\/a>
\nInstall skills via skills.sh<\/a>
\nAdd a skill to Alexa<\/a>
\nMarketer example<\/a>
\nFAQs<\/a>
\nConclusion<\/a><\/div>\n<\/div>\n

<\/p>\n

\n

What \u201cSkills\u201d Mean (and why they\u2019re not just a Claude feature)<\/h2>\n

The word \u201cskill\u201d is overloaded. In most modern assistant ecosystems, an AI skill<\/a> means:
\na packaged capability<\/b> that adds new behavior\u2014often a procedure, a tool connection, or a reusable playbook.<\/p>\n

\n
\n
Agent Skills (skills.sh ecosystem)<\/div>\n
Installable skills for AI agents using a CLI. Great for reusable internal workflows and automation.<\/div>\n
Think: \u201crun my campaign teardown checklist\u201d or \u201cgenerate my reporting summary format.\u201d<\/div>\n<\/div>\n
\n
Voice Skills (Alexa Skills Kit)<\/div>\n
Published voice experiences with invocation names and interaction models. Great for consumer-facing voice utilities.<\/div>\n
Think: \u201cAlexa, open Brand Assistant\u201d or \u201cAlexa, ask Promo Finder.\u201d<\/div>\n<\/div>\n<\/div>\n
\n
Marketing takeaway<\/div>\n
If the goal is internal repeatability (your team running the same workflow every week), skills.sh-style agent skills are a strong fit.
\nIf the goal is a public voice experience (customers talking to your brand), Alexa skills are the right model.<\/div>\n<\/div>\n<\/section>\n

<\/p>\n

\n

Key Statistics on How to Add AI Skills (especially for Alexa skills)<\/h2>\n
\n
\n
\n
Alexa skills available worldwide<\/div>\n
\n
160k+<\/div>\n
catalog size<\/div>\n<\/div>\n
Discovery is competitive<\/div>\n<\/div>\n
\n
Skills with >100 ratings (estimate)<\/div>\n
\n
<1%<\/div>\n
engagement reality<\/div>\n<\/div>\n
Most skills stay niche<\/div>\n<\/div>\n
\n
Skill building toolchain<\/div>\n
\n
3<\/div>\n
paths<\/div>\n<\/div>\n
Console + CLI + APIs<\/div>\n<\/div>\n<\/div>\n
Marketer takeaway: If you\u2019re building an Alexa skill, plan for distribution<\/b> (invocation memorability, retention, ratings).
\nIf you\u2019re building internal agent skills, plan for repeatability<\/b> (checklists, consistent output format, handoff).<\/div>\n
Sources: The Verge on \u201c160,000+ skills,\u201d Voicebot estimate on ratings concentration, Amazon Alexa documentation on build tools.<\/div>\n<\/div>\n<\/section>\n

<\/p>\n

\n

A Simple Decision Flow: Should You Add an Agent Skill or Build an Alexa Skill?<\/h2>\n

<\/p>\n

\n
Decision flow (marketer-friendly)<\/div>\n
\n
\n
Question 1<\/div>\n
Who is the user?<\/div>\n
Internal team or public customers?<\/div>\n<\/div>\n
\n
\u2192<\/div>\n<\/div>\n
\n
If internal<\/div>\n
Use agent skills<\/div>\n
Standardize workflows: briefs, teardowns, reports, QA checklists.<\/div>\n<\/div>\n
\n
\u2198<\/div>\n<\/div>\n
\n
If public<\/div>\n
Build an Alexa skill<\/div>\n
Voice-first experience: invocation, intents, hosting, certification, publishing.<\/div>\n<\/div>\n<\/div>\n
Shortcut: If the \u201cskill\u201d is basically a checklist + output format, start with agent skills. If it needs voice interaction and public distribution, go Alexa.<\/div>\n<\/div>\n<\/section>\n

<\/p>\n

\n

How to Add AI Skills to Claude (Claude Code)<\/h2>\n

\"How<\/p>\n

In Claude Code, skills extend Claude\u2019s capabilities using a simple file-based pattern.
\nClaude Code can load skills automatically when relevant, and you can also invoke them directly as slash commands.<\/p>\n

<\/p>\n

\n
\n
Step 1: Create a skill folder<\/div>\n
Use the Claude Code convention (example): .claude\/skills\/your-skill\/<\/b>. You can select from the top AI skills<\/a> from the list.<\/div>\n<\/div>\n
\n
Step 2: Add a SKILL.md file<\/div>\n
Put instructions in SKILL.md<\/b>. This is your playbook: inputs \u2192 steps \u2192 checks \u2192 output format. You can select from the top AI skills from the list.<\/div>\n<\/div>\n
\n
Step 3: Invoke it as a slash command<\/div>\n
Claude Code skills can be invoked like \/skill-name<\/b> when you want to run that workflow.<\/div>\n<\/div>\n<\/div>\n
\n
Marketer tip: make your SKILL.md enforce \u201cproof blocks\u201d<\/div>\n
Example rule: \u201cInclude 3 statistics with sources + 2 competitor examples + a CTA checklist.\u201d
\nThis is how you stop AI content from becoming generic.<\/div>\n<\/div>\n<\/section>\n

<\/p>\n

\n

How to Install Skills Using skills.sh (the open agent skills ecosystem)<\/h2>\n

If you don\u2019t want to write your own skill from scratch, skills.sh provides a directory + CLI for installing skills into supported agents.
\nThe CLI can run via npx<\/b> (no manual install required).<\/p>\n

<\/p>\n

\n
Install a skill (typical pattern)<\/div>\n
\n
npx skills add <owner\/repo><\/div>\n
npx skills add vercel-labs\/agent-skills<\/div>\n<\/div>\n
After installation, your agent can use the skill when relevant\u2014or you can explicitly run it depending on the agent\u2019s interface.<\/div>\n<\/div>\n
\n
Where this helps marketers most<\/div>\n
Build a small internal \u201cskill library\u201d for repeatable tasks: campaign teardown, landing page audit, SEO refresh<\/a> checklist, creative testing plan, weekly KPI narrative.<\/div>\n<\/div>\n<\/section>\n

<\/p>\n

\n

How to Add AI Skills in Alexa: The Practical Steps (developer console)<\/h2>\n

\"How<\/p>\n

Alexa skills are built and published via the Alexa Skills Kit (ASK).
\nYou typically use the Alexa developer console (GUI), or the ASK CLI \/ Skill Management API for automation<\/a>.<\/p>\n

<\/p>\n

\n
\n
Step 1: Create a new skill in the Alexa developer console<\/div>\n
Choose the experience type, a model (interaction style), and primary locale (language\/country).<\/div>\n<\/div>\n
\n
Step 2: Define your interaction model<\/div>\n
Set an invocation name and intents. Keep it simple: one clear \u201cjob\u201d for the user, not 20 features.<\/div>\n<\/div>\n
\n
Step 3: Connect logic (hosting\/service)<\/div>\n
Use the hosting option that fits your team (common paths include AWS-backed hosting or API-based approaches).<\/div>\n<\/div>\n
\n
Step 4: Test, validate, and publish<\/div>\n
Iterate with testing tools, then submit for certification and publishing.<\/div>\n<\/div>\n<\/div>\n
\n
Important reality check<\/div>\n
With a huge catalog of Alexa skills, distribution and retention matter.
\nTreat the skill like a product: a single job, a memorable invocation name, and a reason to return.<\/div>\n<\/div>\n<\/section>\n

<\/p>\n

\n

A Marketer-Ready Example Skill: \u201cCompetitor Ad Teardown\u201d (weekly)<\/h2>\n

If you want a high-ROI first skill for marketers<\/a>, don\u2019t start with \u201cwrite blog posts.\u201d
\nStart with a workflow that improves everything you ship: competitor teardown \u2192 angle extraction \u2192 test plan.<\/p>\n

<\/p>\n

\n
\n
\n
Inputs<\/div>\n
    \n
  • 3\u20135 competitor ads (screenshots or links)<\/li>\n
  • Landing page URL(s)<\/li>\n
  • Your offer + audience<\/li>\n<\/ul>\n<\/div>\n
    \n
    Skill steps (the playbook)<\/div>\n
      \n
    • Tag hooks by \u201cangle\u201d (pain, outcome, proof, offer)<\/li>\n
    • Extract proof elements (numbers, badges, guarantees)<\/li>\n
    • Generate 5 testable variants (format + CTA)<\/li>\n<\/ul>\n<\/div>\n
      \n
      Outputs (what \u201cgood\u201d looks like)<\/div>\n
        \n
      • A 1-page teardown summary<\/li>\n
      • A\/B test matrix (hypothesis, KPI, duration)<\/li>\n
      • Creative spec cards for designers<\/li>\n<\/ul>\n<\/div>\n<\/div>\n
        If you standardize this teardown as a skill, every marketer on the team can run it and produce consistent outputs\u2014no \u201cit depends\u201d formatting.<\/div>\n<\/div>\n<\/section>\n

        <\/p>\n

        \n

        FAQs: How to Add AI Skills<\/h2>\n
        \n
        \nIs \u201cadding a skill to Claude\u201d the same as \u201cadding a skill to Alexa\u201d?<\/summary>\n
        No. Claude Code skills are file-based playbooks used by an agent (often internal workflows). Alexa skills are published voice experiences with interaction models and distribution considerations.<\/div>\n<\/details>\n
        \nWhat\u2019s the fastest \u201cfirst skill\u201d for a marketing team?<\/summary>\n
        Start with a weekly workflow: competitor teardown, landing page audit, or performance reporting narrative. These create compounding benefits across every campaign.<\/div>\n<\/details>\n
        \nHow does skills.sh help if I\u2019m not a developer?<\/summary>\n
        It provides installable skills via a CLI and a directory of skill packages. Your team can adopt existing skills and refine the playbooks over time.<\/div>\n<\/details>\n<\/div>\n<\/section>\n

        <\/p>\n

        \n

        Conclusion<\/h2>\n

        \u201cSkills\u201d are becoming the default way to extend assistants because they solve the real problem: repeatability<\/b>. If you want internal team automation, package your workflows as agent skills (and consider installing from skills.sh). If you want a public, voice-first product experience, build an Alexa skill\u2014but treat distribution as part of the work. Either way, the win is the same: turn your best marketing procedure into a reusable capability.<\/p>\n<\/section>\n


        \n