Rocky-Project Hail Mary agent skill that cut output tokens ~47%
Persona-based prompting cuts tokens 47% without breaking code like Caveman styles do.

Packages product-thinking into an Agent Skill so your agent can answer like 'Steve Jobs' — heuristics, constraints and example responses live inside a deployable skill compatible with MCP-enabled platforms. Clever and immediately useful for design critiques, but the post prioritizes argument over onboarding: show the skill manifest, install steps and sample inputs up front and this would convert curiosity into actual usage.
Developers and product designers building with AI agents, makers who want persona-based guidance
I found it very useful recently to have my own "Steve Jobs in my pocket" when developing with Claude Code. I can ask it what Steve Jobs would think and usually comes up with great user, simplicity, and story-first thinking.
I thought this might be useful for others, so sharing it here. In the article I also describe why I think skills are the new apps, just with different economics. If you haven't tried them yet, it's certainly worth it. Much more of a change to "default" Claude Code than without.
Happy to hear any feedback!
Persona-based prompting cuts tokens 47% without breaking code like Caveman styles do.
Claude Skill for agent evals, but LangSmith and Arize already own this.
Decentralized skill discovery protocol for agents, like robots.txt for LLM actions.
Clever workflow loop separation, but it's a personal Claude plugin for one methodology.
Finally, Rails conventions for LLM calls instead of scattered API code in controllers.
Curated directory of Claude skills when discovery is the actual bottleneck.