[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-2841":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":15,"subscribersCount":15,"size":15,"stars1d":15,"stars7d":15,"stars30d":16,"stars90d":15,"forks30d":15,"starsTrendScore":15,"compositeScore":17,"rankGlobal":10,"rankLanguage":10,"license":18,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":21,"hasPages":19,"topics":22,"createdAt":10,"pushedAt":10,"updatedAt":31,"readmeContent":32,"aiSummary":33,"trendingCount":15,"starSnapshotCount":15,"syncStatus":34,"lastSyncTime":35,"discoverSource":36},2841,"Prompt-sensei","chengzhongwei\u002FPrompt-sensei","chengzhongwei","Local-first prompt coach for Claude Code and Codex that improves prompts, observes prompting habits, and analyzes local history with explicit consent.","",null,"JavaScript",117,12,9,0,3,43.64,"Apache License 2.0",false,"main",true,[23,24,25,26,27,28,29,30],"ai-coding","claude-code","claude-code-skills","claude-skills","codex","developer-tools","local-first","prompt-engineering","2026-06-12 04:00:16","# Prompt Sensei\n\n> Prompting is becoming one of the most important engineering skills in the AI era. Prompt Sensei helps you improve your prompts while teaching you how to improve.\n\n[中文说明](README-zh.md)\n\n[Quickstart](#5-minute-quickstart) · [FAQ](docs\u002Ffaq.md) · [Privacy](docs\u002Fprivacy.md) · [Examples](examples\u002Fprompt-gallery.md) · [Advanced setup](docs\u002Fadvanced-setup.md) · [Discussions](https:\u002F\u002Fgithub.com\u002Fchengzhongwei\u002FPrompt-sensei\u002Fdiscussions)\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"assets\u002Fprompt-sensei-banner-en.jpg\" alt=\"Prompt Sensei banner: Claude Code and Codex support, local-first optional hooks, no raw prompts stored, and stage-aware feedback.\" \u002F>\n\u003C\u002Fp>\n\nPrompt Sensei is a quiet, local-first prompt coach for Claude Code and Codex. It gives stage-aware feedback, rewrites rough prompts into better ones, looks back on local history when you ask, and helps you practice one habit at a time.\n\nIt is built as a real host-integrated skill, not just a markdown rubric: Prompt Sensei supports hook-based observe comments, opt-in auto-observe hooks, hash-only prompt captures, quiet Stop-hook persistence, and compact-safe continuity where the host exposes the lifecycle events.\n\nNo cloud. No telemetry. No leaderboard. No raw prompt archive.\n\nPrompt Sensei's score is a coaching signal, not an objective grade or a guarantee of better output. The real test is whether the improved prompt produces a more useful first draft, fewer clarification turns, safer agent behavior, and easier verification.\n\n---\n\n## Why This Exists\n\nAI coding agents do real work: they edit files, run commands, and make implementation choices. A vague prompt does not just produce a vague answer; it can produce broad edits, hidden assumptions, and extra revision loops.\n\nPrompt Sensei is built around a simple belief:\n\n- Prompting is an engineering skill.\n- Engineering skills improve through feedback.\n- Feedback should be kind, specific, private, and useful.\n\nIt does not treat `fix this test` as a bad prompt. It treats it as an early-stage prompt, then teaches the next step.\n\n---\n\n## 5-Minute Quickstart\n\nInstall for Claude Code:\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fchengzhongwei\u002FPrompt-sensei ~\u002F.claude\u002Fskills\u002Fprompt-sensei\n(cd ~\u002F.claude\u002Fskills\u002Fprompt-sensei && npm install && npm run build)\n```\n\nThen start live coaching with `\u002Fprompt-sensei observe` inside Claude Code. For optional auto-start and privacy settings, run `\u002Fprompt-sensei setup`.\n\nInstall for Codex:\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fchengzhongwei\u002FPrompt-sensei ~\u002F.codex\u002Fskills\u002Fprompt-sensei\n(cd ~\u002F.codex\u002Fskills\u002Fprompt-sensei && npm install && npm run build)\n```\n\nThen ask Codex in natural language: `Use prompt-sensei observe mode.` For optional hook-based auto-start, run `npm run setup-hooks -- auto-observe folder` or use `\u002Fprompt-sensei setup` after the skill is loaded.\n\nStart live coaching in Claude Code:\n\n```txt\n\u002Fprompt-sensei observe\n```\n\nTry improving a rough prompt:\n\n```txt\n\u002Fprompt-sensei improve \"fix this test\"\n```\n\nExample output:\n\n```txt\nPrompt Sensei Improve\n=====================\nStage:    Exploration\nScore:    70 \u002F 100  (Good)\n\nWhat is missing:\n  - failing test name\n  - expected behavior\n  - actual error output\n\nImproved prompt:\n  Help me debug this failing test.\n\n  Test: [test name]\n  Expected: [what should happen]\n  Actual: [error output or wrong behavior]\n  Related file: [file path]\n\n  Return:\n  1. Likely root cause\n  2. Minimal fix\n  3. Test command to verify\n\nHabit to practice next:\n  Add expected and actual behavior before asking for a fix.\n```\n\nSee [examples\u002Fprompt-gallery.md](examples\u002Fprompt-gallery.md) for more copyable before\u002Fafter prompts.\n\n---\n\n## Supported Environments\n\nPrompt Sensei works best when the host tool can load the skill directly. It supports both Claude Code and Codex, with host-native hooks for optional auto-observe.\n\n| Invocation style | Environments |\n|---|---|\n| Direct skill command, e.g. `\u002Fprompt-sensei improve \"fix this test\"` | Claude Code |\n| Natural language, e.g. `Use prompt-sensei to improve this prompt...` | Codex |\n\nCursor and other AI coding tools are not supported yet. `\u002Fprompt-sensei observe` is a Claude Code skill command. In Codex, natural-language requests depend on the host agent loading the installed skill. See [advanced setup](docs\u002Fadvanced-setup.md#host-support) for the Claude Code\u002FCodex hook matrix and trust notes.\n\n---\n\n## Beginner Path\n\nStart with:\n\n```txt\n\u002Fprompt-sensei observe\n```\n\nOn first use, Prompt Sensei asks for local storage consent, starts coaching, then offers an optional auto-start choice. The simplest path is manual start; you can still use `\u002Fprompt-sensei observe` whenever you want.\n\nFor a one-time guided setup, use:\n\n```txt\n\u002Fprompt-sensei setup\n```\n\nSetup covers observe consent, optional host auto-start hooks, and whether to save redacted prompt previews. In Codex, a trust prompt for new or changed command hooks is expected; inspect commands with `\u002Fhooks` before enabling them. Advanced details live in [docs\u002Fadvanced-setup.md](docs\u002Fadvanced-setup.md).\n\n---\n\n## Commands\n\n```txt\n\u002Fprompt-sensei [observe|improve|lookback|setup|help]\n```\n\nEssential commands:\n\n```txt\n\u002Fprompt-sensei observe              # start live coaching\n\u002Fprompt-sensei improve \"fix this\"   # rewrite a prompt with one teaching note\n\u002Fprompt-sensei lookback             # analyze selected local prompt history\n\u002Fprompt-sensei setup                # guided setup\n\u002Fprompt-sensei help\n```\n\nMore commands:\n\n```txt\n\u002Fprompt-sensei stop                 # stop coaching this session\n\u002Fprompt-sensei report               # show local session trends\n\u002Fprompt-sensei settings             # show local settings\n\u002Fprompt-sensei update               # pull latest version and rebuild\n\u002Fprompt-sensei clear                # delete local Prompt Sensei data\n```\n\nIn Codex, use natural language equivalents such as `Use prompt-sensei observe mode.` or `Use prompt-sensei to improve this prompt: \"fix this test\"`.\n\nSee [docs\u002Fadvanced-setup.md](docs\u002Fadvanced-setup.md) for settings commands, hook setup, Codex examples, and consent scopes.\n\nFor direct script checks, see [advanced setup](docs\u002Fadvanced-setup.md#local-script-checks).\n\n---\n\n## Settings\n\nPrompt Sensei stores local preferences in `~\u002F.prompt-sensei\u002Fsettings.json`. Defaults are intentionally quiet: auto observe off, redacted prompt previews off, and raw prompts never stored. Use `\u002Fprompt-sensei setup` for a guided path or [docs\u002Fadvanced-setup.md](docs\u002Fadvanced-setup.md) for the full settings reference.\n\n---\n\n## How It Teaches\n\nPrompt Sensei is stage-aware. A short exploration prompt can be reasonable early, while an execution prompt needs clearer context, boundaries, and verification.\n\n| Stage | Meaning | Example |\n|---|---|---|\n| Exploration | Still figuring out the problem | `why is this broken` |\n| Diagnosis | Have evidence or symptoms | `expected \u002Flogin, actual \u002Fdashboard` |\n| Execution | Want implementation or changes | `implement this with these constraints` |\n| Verification | Want correctness checks | `find edge cases and test commands` |\n| Reusable workflow | Want a repeatable process | `create a code review checklist` |\n| Action | Short follow-through directive | `ok commit and push to main` |\n\nThe coaching line stays small:\n\n> **[[Sensei: 68\u002F100 · Diagnosis; Tip: add the error message and file path]]()**\n\nThink of the score as prompt readiness for the current stage, not universal prompt quality. A 100\u002F100 prompt can still produce a weak answer if the model lacks domain knowledge, the task is ambiguous outside the prompt, or the rubric does not fit the user's domain.\n\nThe report focuses on repeated habits:\n\n```txt\nNext habit:      End with the command, test, or edge case that proves the work.\nRepeated gap:    add-verification-command (5×)\nAverage score:   81 \u002F 100  (Good)\n```\n\nFor the full philosophy, read [docs\u002Fphilosophy.md](docs\u002Fphilosophy.md). For scoring details, read [docs\u002Fscoring-rubric.md](docs\u002Fscoring-rubric.md).\n\n---\n\n## How Do I Know It Helps?\n\nPrompt Sensei does not prove that a prompt is better just because the number went up. It gives a structured coaching signal.\n\nTo check whether it helped, compare the original and improved prompt on the same task: fewer clarification turns, fewer surprising edits, a better first answer, and easier verification. For calibration details and the skeptical version, see [FAQ](docs\u002Ffaq.md).\n\n---\n\n## Lookback\n\n`\u002Fprompt-sensei lookback` analyzes selected local Claude Code or Codex history after separate consent. It can inspect one session or recent sessions, generate coaching, and save a Markdown report only after confirmation. Privacy details live in [docs\u002Fprivacy.md](docs\u002Fprivacy.md#lookback-privacy), and workflow details live in [docs\u002Fskill-flows.md](docs\u002Fskill-flows.md#lookback-flow).\n\n---\n\n## Optional Host Hooks\n\nHost hooks provide opt-in auto-start, hash-only background captures, hook-triggered observe comments, and quiet persistence of the final Sensei score line. They are optional and stay quiet unless consent\u002Fsettings allow them. See [advanced setup](docs\u002Fadvanced-setup.md#host-support), [Claude example hooks](examples\u002Fclaude-settings.example.json), and [Codex example hooks](examples\u002Fcodex-hooks.example.json).\n\n---\n\n## Privacy\n\nPrompt content is sensitive. Prompt Sensei stores nothing until you consent. After consent, it stores local metadata under `~\u002F.prompt-sensei\u002F`; raw prompts are never stored.\n\nPrompt Sensei's local scripts do not send prompt text, scores, reports, or local event data to a service. Lookback may show redacted user prompts to the current AI agent after separate consent.\n\nSee [docs\u002Fprivacy.md](docs\u002Fprivacy.md) for details.\n\n---\n\n## Contributing\n\nGood first areas:\n\n- realistic prompt improvement examples\n- scoring rubric improvements\n- redaction rule improvements\n- report improvements\n- support for other AI coding tools\n\nPlease keep changes aligned with the core philosophy: quiet, local-first, encouraging, and privacy-aware.\n\n---\n\n## License\n\nApache-2.0 — Copyright 2026 Chengzhong Wei\n","Prompt Sensei 是一个针对 Claude Code 和 Codex 的本地优先提示教练工具，旨在帮助用户改进提示、观察提示习惯并分析本地历史记录。该项目采用 JavaScript 编写，具备阶段感知反馈、自动优化粗糙提示以及基于用户同意的历史回顾等核心功能。它通过集成到主机环境中的技能实现，不依赖云端存储或遥测数据，确保了用户的隐私安全。Prompt Sensei 适用于需要提高AI编码助手如Claude Code和Codex使用效率的开发者，特别是在希望提升自己编写有效提示能力的情况下。",2,"2026-06-11 02:51:18","CREATED_QUERY"]