Skill vs Sub-Agent vs MCP Explained

Today I break down a big news item I think is flying under the radar: OpenAI quietly launched Skills for Codex, and I explain what that means (and how it differs from sub-agents and MCPs). I then share a fast-moving trend I’m watching and why it’s a strong wedge for a simple app. After that, I recommend the to-do app I’ve used for 14 years and give away a startup idea. I close with a practical 6-step framework for going from idea → viral validation → mobile app launch in 2026.

Timestamps

00:00 – Intro: the new format (news, trend, app, startup idea, framework)

00:40 – AI New Item: OpenAI launches Skills for Codex

05:45 – Trend: Face Yoga

07:56 – App Recommendation: Things

09:33 – Startup Idea: Call-an-expert service for non-developers stuck at 80% done

14:44 – Framework: Viral Mobile App Framework

Key Points

• OpenAI “Skills” make Codex/ChatGPT more reusable and consistent by packaging repeatable workflows.

• A “skill” is the recipe, a “sub-agent” is extra worker instances, and an “MCP” is the tool access plug.

• Face yoga is an emerging sub-niche with clear app potential (simple routines, monetization via paid or ads).

• Last 20 is a practical marketplace idea: pay for 15 minutes of expert unblock help to finish the last 20%.

• Viral validation favors apps that are visually obvious, explainable in three words, and tied to insecurity-driven outcomes.

Numbered Section Summaries

1. OpenAI Skills: The Quiet Upgrade I walk through OpenAI’s launch of Skills for Codex—reusable bundles of instructions/scripts/resources that can be called directly or chosen automatically. I’m excited because this makes agent workflows more consistent and scalable across tasks.

2. The Foundation: Skill vs Sub-Agent vs MCP I clarify the taxonomy: a skill is the written playbook, sub-agents are extra “worker” copies of the model that split a big job, and MCPs are what let the model access external systems like tickets or repos. This is the mental model I want everyone using going into 2026.

3. The Trend: Face Yoga As An App Wedge I share a niche trend I’m seeing—face yoga—and why it’s a product opportunity similar to how yoga apps became huge. I call out the obvious app angles: guided routines, jawline/face-slimming programs, and content-driven growth via short videos.

4. The Tool: Things (My Simple Focus System) I recommend the Things to-do app because it’s simple: “Today,” “Upcoming,” and “Someday,” without a monthly fee. I also note what’s missing (I’d like more AI features), but it still wins for focus if you don’t want a “kitchen sink” system.

5. The Startup Idea: Last 20 (Phone-A-Friend For Vibe Coders) I give away the idea: builders get stuck at 80% after using Cursor/Replit/V0, so Last 20 matches them with someone who’s solved that exact wall before. The product is a fast screen-share session—problem solved—priced per session or bundled for teams/agencies, with the marketplace taking a cut.

6. The Distribution Framework: Viral Validation → Launch I share a 6-step process: warm up the account, design a visually obvious app, build a tiny MVP fast, post daily until something hits, build the community before the product, then launch with a hard paywall and keep content rolling. It’s a simple playbook for getting to organic traction in 2026.

The #1 tool to find startup ideas/trends - https://www.ideabrowser.com

LCA helps Fortune 500s and fast-growing startups build their future - from Warner Music to Fortnite to Dropbox. We turn 'what if' into reality with AI, apps, and next-gen products https://latecheckout.agency/

The Vibe Marketer - Resources for people into vibe marketing/marketing with AI: https://www.thevibemarketer.com/

FIND ME ON SOCIAL

X/Twitter: https://twitter.com/gregisenberg

Instagram: https://instagram.com/gregisenberg/

LinkedIn: https://www.linkedin.com/in/gisenberg/