{"skill":{"slug":"lobster-agentic-engineering","displayName":"Lobster Agentic Engineering","summary":"Operate as an agentic engineer using eval-first execution, decomposition, and cost-aware model routing.","tags":{"latest":"1.0.0"},"stats":{"comments":0,"downloads":129,"installsAllTime":1,"installsCurrent":1,"stars":0,"versions":1},"createdAt":1775301788654,"updatedAt":1775324506873},"latestVersion":{"version":"1.0.0","createdAt":1775301788654,"changelog":"Initial release of lobster-agentic-engineering skill.\n\n- Enables agentic engineering workflows using eval-first execution and cost-aware model routing.\n- Provides principles for decomposing work into independently-verifiable, risk-scoped units.\n- Details best practices for routing tasks to appropriate AI model tiers (Haiku, Sonnet, Opus).\n- Introduces an eval-first feedback loop with baseline comparisons and regression checks.\n- Recommends strategies for session management and disciplined code review of AI-generated outputs.\n- Emphasizes tracking resource usage and escalating model sophistication only as needed.","license":"MIT-0"},"metadata":null,"owner":{"handle":"wangxiaofei860208-source","userId":"s17937syhc72vf48t0a5914v8d847evs","displayName":"wangxiaofei860208-source","image":"https://avatars.githubusercontent.com/u/260913269?v=4"},"moderation":null}