[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-78236":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":5,"homepage":8,"htmlUrl":8,"language":8,"languages":8,"totalLinesOfCode":8,"stars":9,"forks":10,"watchers":11,"openIssues":12,"contributorsCount":12,"subscribersCount":12,"size":12,"stars1d":12,"stars7d":13,"stars30d":14,"stars90d":12,"forks30d":12,"starsTrendScore":11,"compositeScore":15,"rankGlobal":8,"rankLanguage":8,"license":8,"archived":16,"fork":16,"defaultBranch":17,"hasWiki":18,"hasPages":16,"topics":19,"createdAt":8,"pushedAt":8,"updatedAt":20,"readmeContent":21,"aiSummary":22,"trendingCount":12,"starSnapshotCount":12,"syncStatus":23,"lastSyncTime":24,"discoverSource":25},78236,"boris-prompts","LingyiChen-AI\u002Fboris-prompts","LingyiChen-AI",null,153,16,1,0,5,47,3.69,false,"main",true,[],"2026-06-12 02:03:46","# boris-prompts\n\nAn agent skill that writes high-quality prompts for any LLM (Claude, Claude Code, ChatGPT, Gemini, Cursor, Windsurf, etc.) using **Boris's methodology** — the patterns Boris (Anthropic, Claude Code's creator) shared in his \"Pro Tips & Tricks\" talk.\n\nIf your request is unclear, the skill asks 1–3 targeted clarifying questions before writing. The output is short, copy-pasteable, and follows the five principles below.\n\n## Original tweet\nhttps:\u002F\u002Fx.com\u002FEtudecn\u002Fstatus\u002F2057238154701426726\n\n## Install\n\nThis skill is distributed via the [`skills`](https:\u002F\u002Fgithub.com\u002Fvercel-labs\u002Fskills) CLI (`npx skills`). Works with Claude Code, Codex, Cursor, OpenCode, and 50+ other agents.\n\n```bash\n# GitHub shorthand\nnpx skills add LingyiChen-AI\u002Fboris-prompts --skill boris-prompts\n\n# Or full URL\nnpx skills add https:\u002F\u002Fgithub.com\u002FLingyiChen-AI\u002Fboris-prompts --skill boris-prompts\n\n# Or point at the skill folder directly\nnpx skills add https:\u002F\u002Fgithub.com\u002FLingyiChen-AI\u002Fboris-prompts\u002Ftree\u002Fmain\u002Fskills\u002Fboris-prompts\n\n# Install globally (available across all projects)\nnpx skills add LingyiChen-AI\u002Fboris-prompts --skill boris-prompts -g\n\n# Install only for Claude Code\nnpx skills add LingyiChen-AI\u002Fboris-prompts --skill boris-prompts -a claude-code\n```\n\nAfter install, restart your agent (or start a new session) and the skill loads automatically.\n\n## What it does\n\nWhen you say things like:\n\n- `write a prompt that makes Claude convert all our logs to JSON`\n- `how should I prompt Cursor to add dark mode`\n- `this ChatGPT prompt isn't working — keeps going off the rails`\n- `give me a good way to ask Gemini to refactor this module`\n\n…the skill triggers. If it has enough info, it writes the prompt directly. If not, it asks at most three questions (via `AskUserQuestion`) and then writes it.\n\n## The five principles\n\nThe skill applies these in order:\n\n1. **Short beats long.** A two-sentence prompt usually beats a screen-filling one. If you find yourself writing more than five lines, half of it probably belongs in a persistent-context file.\n2. **\"Make a plan first\" is the single highest-ROI addition.** Appending *\"Before you write code, make a plan and run it by me for approval\"* upgrades almost any non-trivial prompt.\n3. **Don't spec every detail.** Point the model at the right starting place (\"look at how `X` is implemented\") instead of re-describing the codebase.\n4. **A feedback loop beats detailed instructions.** Give the model a way to verify (tests, screenshots, lint) and let it iterate, rather than spelling out every step.\n5. **Persistent context lives in files \u002F settings, not the prompt.** `CLAUDE.md`, `.cursorrules`, Custom Instructions, Projects, Gems — whichever the target supports.\n\n## Example\n\n**You:** \"I want Claude to add JWT auth to my Express app. There are tests.\"\n\n**Skill output:**\n\n```\nAdd JWT auth to the Express app. Look at how existing middleware is structured first, then make a plan and run it by me. After implementing, run the test suite and iterate until it passes.\n```\n\n> Plan-first + feedback loop applied. Anchors Claude in existing patterns rather than re-describing them.\n\n## Repo layout\n\nThis repo follows the convention the `skills` CLI expects: a top-level `skills\u002F` directory containing one folder per skill, each with a `SKILL.md`.\n\n```\nboris-prompts\u002F                  (this repo)\n└── skills\u002F\n    └── boris-prompts\u002F          (the skill)\n        └── SKILL.md\n```\n\nTo add more skills later, drop them in as siblings under `skills\u002F`.\n\n## Credits\n\nMethodology from Boris's \"Claude Code Pro Tips\" talk (Anthropic). This repo just packages it as an agent skill.\n\nCommunity: \u003Chttps:\u002F\u002Flinux.do>\n\n## License\n\nMIT\n","boris-prompts 是一个能够为多种语言模型（如Claude、ChatGPT等）生成高质量提示词的技能。它基于Boris方法论，通过分析用户需求并提出最多三个针对性问题来确保理解准确后再生成简洁可复制的提示文本。遵循五大原则：短胜于长、先规划后执行、不需详尽描述细节、反馈循环优于详细指令、持久上下文应存于文件而非提示中。适用于需要优化与大型语言模型交互质量的场景，比如代码开发辅助、文档生成等任务。通过`skills` CLI工具安装即可轻松集成到支持的平台上使用。",2,"2026-06-11 03:56:38","CREATED_QUERY"]