[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-83158":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":8,"languages":8,"totalLinesOfCode":8,"stars":9,"forks":10,"watchers":11,"openIssues":10,"contributorsCount":10,"subscribersCount":10,"size":10,"stars1d":10,"stars7d":12,"stars30d":12,"stars90d":10,"forks30d":10,"starsTrendScore":10,"compositeScore":13,"rankGlobal":8,"rankLanguage":8,"license":8,"archived":14,"fork":14,"defaultBranch":15,"hasWiki":16,"hasPages":14,"topics":17,"createdAt":8,"pushedAt":8,"updatedAt":18,"readmeContent":19,"aiSummary":20,"trendingCount":10,"starSnapshotCount":10,"syncStatus":21,"lastSyncTime":22,"discoverSource":23},83158,"the-first-principle-of-agi","yanjin101\u002Fthe-first-principle-of-agi","yanjin101",null,99,0,96,3,35.8,false,"master",true,[],"2026-06-12 04:01:40","my first principle of agi\n\ni don't actually recommend u use this principle without fully understanding. it's just a reference at what's possible\n\n# The First Principle of AGI\n\nAny continuous task can be broken down into discrete, small tasks that an LLM can complete without error, achieved through multi-step execution.\n\n> **Analogy to the human brain:** rational logic (task decomposition) vs. intuitive perception (LLM completing error-free small tasks)\n\n---\n\n## How Does an Agent Learn Tasks?\n\nPreserve the prompt experience of human pioneers, compressed into `skill.md` through summarization or hard-coded into code.\n\n> **Analogy to the human brain:** sparse neural connections (summarized into `skill.md`), dense neural connections (hard-coded)\n\n---\n\n## How Does an Agent Decompose Tasks?\n\nThrough human pioneers refining the task logic, exhaustively enumerating all cases at the granularity level that an LLM can reliably complete.\n\n---\n\n## How Does an Agent Scale?\n\nContinuously persist task experience from all industries and professions to disk, expanding the granularity of the minimum task an LLM can reliably complete.\n\n---\n\n> Implement this guiding ideology and become the AI Adventists.\n","该项目提出了一个关于通用人工智能（AGI）的基本原则，强调任何连续任务都可以被分解为一系列小的、无错误的任务，通过多步骤执行来完成。其核心功能在于提供了一种方法论，通过人类先驱者的经验总结或硬编码方式，使代理能够学习和分解任务，并且随着跨行业实践经验的积累不断扩展任务处理能力。适合于对探索AI在复杂任务中的应用感兴趣的研究者以及希望深入了解如何构建更高效的人工智能系统的开发者参考使用。",2,"2026-06-11 04:10:18","CREATED_QUERY"]