AI Plugin Economy: The New Aftermarket Boom

The Pattern

Jason Cohen’s MicroConf talk on designing the ideal bootstrapped business identified the aftermarket plugin as the most reliable bootstrap pattern decades ago: pick a platform with a massive user base, find a pain point, build a plugin. WordPress hosting (WP Engine), Photoshop plugins (Alien Skin), QuickBooks connectors — the pattern has generated billion-dollar outcomes from what started as side projects.

Now the same pattern is playing out at unprecedented speed across AI tools. Claude has skills. Cursor has rules and a community directory pulling $35K/month. n8n has workflow templates. Every MCP server is a plugin. Every system prompt is a configurable add-on. The AI plugin economy is not a prediction — it is already here, and the early movers are already monetizing.

The structural difference from the WordPress era is velocity. A WordPress plugin took weeks to build. A Claude skill or Cursor rule can be written in an afternoon. An n8n workflow template can be exported and packaged in hours. The barrier to creating plugins has collapsed, which means the value shifts from building the plugin itself to curating, packaging, and distributing quality plugins at scale.

Key Quotes

“After-market tools — pick a platform with a huge following, solve their pain. WordPress hosting, Photoshop plugins, QuickBooks ODBC connector since 1984. Keywords obvious, customers findable, problem clear.” — Jason Cohen, MicroConf, 37:34

Cohen’s framework is 15+ years old and still the most precise articulation of why this pattern works. The QuickBooks ODBC connector has been a viable business since 1984 — over 40 years of revenue from a single plugin. The AI equivalents are just beginning their run.

“Claude skills are not just for Claude. The same pattern — a structured set of instructions and tools packaged for reuse — applies to any AI coding agent. This is the new plugin format.” — Cole Medin, 14:54

Medin’s insight is that Claude skills are a portable concept. A skill is just a well-structured system prompt plus tool configuration plus context. That same bundle works in Cursor, Windsurf, Copilot, or any other AI coding tool. The plugin format is converging across platforms, which means building once and distributing everywhere.

“My smartest friends all have one thing in common — they’ve built a library of Claude Code skills for their specific workflows. Not generic productivity tips. Hyper-specific skills for their exact job.” — Greg Isenberg, 0:00

“The problem you intend to solve is so much more important than almost anything else. Don’t tell me your idea. Tell me the problem it solves and for whom.” — Rob Walling, MicroConf, 1:11

Walling’s advice maps directly onto the AI plugin opportunity. The winning plugins are not “general AI productivity enhancers.” They solve specific problems for specific users: a Claude skill that generates compliant medical documentation, a Cursor rule set optimized for Astro projects, an n8n workflow template that automates invoice processing for freelancers.

Prediction Check

• AI plugin marketplaces emerging: Confirmed. cursor.directory hit $35K/month as a community-driven directory of Cursor rules. Claude’s skill system is growing. MCP server registries are multiplying. The marketplace layer is forming organically.
• Monetization of AI configurations: Early but confirmed. Premium Cursor rules, paid Claude skill bundles, and n8n template packs are all generating revenue. The willingness to pay is strongest among professional developers and agencies who value curated, tested configurations over DIY.
• Platform risk remains manageable: Holds true so far. AI platforms want ecosystems. Anthropic actively promotes skills. Cursor benefits from cursor.directory driving adoption. n8n’s open-source model makes ecosystem lock-in less likely. Cohen’s original observation — platforms want third-party developers solving the long tail — applies even more strongly in AI.

Concrete Ideas

• AI skills marketplace — a curated directory where developers publish and sell Claude skills, Cursor rules, and MCP server configurations. Free tier for open-source plugins, premium listings at $9-49 for professional/commercial-grade ones. Revenue from listing fees plus a 20% cut on sales. The cursor.directory success proves demand; expand the model across all AI coding tools.
• Industry-specific Claude skill packs — bundles of 10-20 Claude skills tailored for specific professions. A pack for healthcare compliance ($49), a pack for legal document drafting ($79), a pack for financial analysis ($49). Each skill is tested, documented, and includes example workflows. Sell direct via Gumroad or through the skill marketplace.
• n8n template studio — productize the n8n workflow template pattern that Cohen predicted. Build production-ready automation templates for specific business niches: real estate lead nurturing, ecommerce order processing, content calendar automation. $29-99 per template, $199-499 per vertical bundle. Include video walkthroughs and 30 days of support.
• MCP server builder and registry — a tool that lets non-developers create custom MCP servers from API documentation. Point it at any API, describe the tools you want, and it generates a working MCP server. Charge $19/mo for the builder, list servers in a public registry. The Cole Medin insight that skills are portable across AI tools makes this especially valuable.
• “Cursor rules for X” generator — a tool that analyzes a codebase and generates optimized Cursor rules for that specific project’s stack, patterns, and conventions. Input: a GitHub repo URL. Output: a complete .cursor/rules directory. Sell as a one-time analysis ($19) or subscription ($9/mo) for ongoing rule updates as the codebase evolves.
• AI agent template marketplace — pre-built agent configurations for common business workflows (customer support, sales qualification, content generation). Each template includes the system prompt, tool configurations, testing suite, and deployment instructions. Target AI agencies who need to deliver fast. $99-299 per template.

Analysis

The AI plugin economy has a structural advantage over previous aftermarket booms: the creation cost is near zero. A WordPress plugin required PHP knowledge, testing across themes, and ongoing compatibility maintenance. A Claude skill requires writing good instructions and testing them. This means the supply side will be enormous — which makes curation and quality the differentiating factor.

cursor.directory is the canonical example. It did not build anything complex. It aggregated community-contributed Cursor rules into a searchable directory with quality signals. $35K/month in revenue from what is essentially a well-curated list. The value is not in the individual rules — it is in the discovery, trust, and packaging.

Arvid Kahl’s audience-first philosophy applies perfectly here. Before building a plugin marketplace, embed yourself in the communities where AI tool users congregate. Understand what workflows people are hacking together, what system prompts they are sharing in Discord, what MCP servers they are recommending on Twitter. The marketplace should feel like it was built by a community insider, because it was.

Rob Walling’s method for finding SaaS ideas — lurking in forums, finding problems that are not yet solved — maps directly onto the AI plugin opportunity. Every AI tool subreddit, every Claude Discord channel, every Cursor community is filled with people sharing workarounds and asking “has anyone built a skill/rule/template for X?” Each of those questions is a product waiting to be packaged.

The niche directory pattern that generates $10M/year in traditional markets applies here too. A general “AI plugins” directory competes with the platforms themselves. A directory of “Claude skills for healthcare professionals” or “n8n templates for real estate agents” has no competition and immediately obvious keywords for organic search.

The risk is commoditization. As AI tools get better, the value of simple plugins decreases. A Claude skill that reformats JSON will be obsolete when Claude handles it natively. The durable plugins are the ones that encode domain expertise — compliance knowledge, industry-specific workflows, professional judgment — that the base model cannot replicate from training data alone.

What to Build

Start with a curated directory of Claude skills or Cursor rules for one specific industry or use case. Not a general-purpose marketplace. Pick a vertical you know (legal, healthcare, ecommerce, content creation) and build the definitive collection of AI tool configurations for that vertical. Contribute 20 high-quality free entries to establish authority. Then add a premium tier ($29-99 for bundles) with tested, documented, production-grade configurations.

The distribution advantage is built into the model. Every free entry ranks in search. Every user who downloads a free skill sees the premium offerings. The platform’s ecosystem drives traffic. You are not competing with the platform — you are extending it, exactly as Cohen prescribed 15 years ago. The only difference is that the platform is now Claude instead of WordPress, and the plugin is a skill file instead of PHP code.