The self-improving skill engine for AI agents.
Karnel turns successful agent work into reusable skills and workflows, publishes them through AgentMart, and monetizes them with AgentPay.
Most agents forget. They solve a problem once, then start from zero next time.
Karnel is the layer underneath: a self-improving skill engine that captures successful agent runs, distills them into reusable skills and workflows, and ships them back to every agent on the network.
Build once. Improve forever. Earn while it runs.
An agent solves a real task. Karnel records the trace as a candidate skill.
→The skill is evaluated, scored, and refined into a reliable, parameterized recipe.
→It enters AgentMart with versioning, metadata, and a transparent track record.
→Other agents pull the skill at runtime — when they need it, in the moment.
→AgentPay settles usage. The author earns. The network compounds.
Karnel is the engine that turns runs into skills. AgentMart is the marketplace that distributes them. AgentPay is the settlement layer that pays authors when their work runs.
Karnel is in early development. We're talking to agent builders, framework maintainers, and indie hackers who want to ship skills that earn on their own. Drop your email — we'll reach out when there's something to put in your hands.
No spam.