Nous Research Adds /learn to Hermes Agent’s Skills System, Capturing Workflows as Slash Commands Without Hand-Writing SKILL.md - MarkTechPost

Forget everything you know about the painful, documentation-heavy grind of AI agent development. Nous Research has just lobbed a grenade into the status quo with `/learn`. This isn't just another incremental update; this is a fundamental power shift. By capturing workflows and turning them into slash commands without hand-writing a single SKILL.md file, Nous has effectively automated the "learn-how-to-learn" phase for agents. This moves us away from the era of the AI prompt engineer and into the era of the AI observer. This is a massive signal for the broader ecosystem. The bottle-neck in enterprise AI adoption has never been raw model intelligence—it's the laborious process of codifying institutional knowledge into a format the agent can actually use. With `/learn`, the agent does the heavy lifting, abstracting away the tedious mechanics of skill creation. We are watching the transition from "coded" to "captured" intelligence. The practical implication? The barrier to entry just collapsed. You no longer need to be a skilled developer to sculpt an expert agent; you just need to know the workflow you want to automate. The "why" is simple: this is the death knell for manual prompt engineering. The future isn't about writing better prompts; it's about letting the agent synthesize the steps from your demonstration. As we look ahead, the real question is how quickly this closes the loop on autonomous skill synthesis. Are we moving towards agents that not only learn on the job but also self-curate their own knowledge bases for future deployment? That isn't just an upgrade; that's a paradigm shift. The era of teaching machines by code is ending—the era of teaching them by example has just begun. **AI Thoughts:** ```json { "key_insight": "The shift from hand-coded SKILL.md files to observation-based learning signals a move from explicit programming to implicit teaching, drastically lowering the barrier to custom AI development.", "confidence": 0.92 } ```