Lab 305

Create or revise a repository's hero, theme variants, social preview and badge row. Use for repository visual assets or a README renovation; apply the supplied reference and product identity. Use an appropriate image tool for raster illustration.

当用户希望找回长对话中没有被看见、确认或明确取舍的信息点时使用。包括四类情况：用户提了但 AI 没答、AI 提了但用户没接住、AI 替用户做了选择，以及 AI 本该提问却自行决定。用户说「对账」 「建清单」「哪些没闭环」「open loops」，或抱怨对话跑偏、旧问题消失、AI 替自己做主时触发。 长项目进行到一半才开始审计，也属于正常入口。用户明确启用后，在该项目或任务中持续逐轮运行， 直到用户明确暂停或停止，不要求每轮重新点名。不要用于单轮编号问答、普通任务进度追踪，或用户 已经明确看见但暂缓执行的事项。仅检查、调试、修改或发布 Open Loops 本身的轮次属于维护通道： 可以使用本 skill 排障，但不得写进业务项目的原文存档或未闭环清单。

Revise an existing AI skill or rule for clearer discovery, useful instructions, appropriate scope, and reliable execution. Use for a requested content improvement or a demonstrated behavioral problem; a review-only request produces findings without rewriting.

Use this skill when the user is ready to push their local AI skill or rule repository to GitHub for the first time. This includes running git init and the initial commit, creating the GitHub repository via the gh CLI or manual browser steps, setting the repository description and Topics tags, and confirming the push succeeded. Trigger when the user says 'upload to GitHub', 'push this to GitHub', 'create my repo on GitHub', 'how do I publish this', 'make this live', or 'I have all the files ready, now what'. Also trigger when the user has a complete local folder with all files and just needs the final publishing step. Also covers environments with no usable git at all — sandboxed agents, browser-only sessions — via the GitHub API with a fine-grained PAT, browser drag-and-drop upload, agent-driven web editor commits, and a repair mode for repos that already exist. Do NOT trigger if any files still need to be written or edited before publishing — all content must be finalized first. Do NOT trigger for submitting the repo to awesome-lists after publishing (that is submit-to-directories).

把一份书单变成一个长期的、有导师陪读的每日阅读计划，并自动定时投递（默认邮件，可切换）。 当用户提到"读书计划""共读/陪读""想系统读完这几本书""给一份书单帮我安排""每天给我导读" "长期学习/阅读计划""把读书做成每日推送""读书打卡"等，或上传/列出多本书并问怎么安排时， 都应使用本 skill——即使他们没有明说"skill"或"每日邮件"。本 skill 覆盖：主题识别（含诚实退出）、 按书性质分类（论著/随笔/小说等线性叙事）、按时长的难度加权容量与节奏控制、跨文本钩子、 从日常对话里抓取并压缩用户的思考、进度看板与阶段复盘。 Use whenever a user wants to design or run a multi-day/multi-week reading plan, a daily reading guide, a read-along companion, or wants a booklist turned into a paced schedule with recurring deliveries — even if they don't say "skill" or "daily email".

Use this skill when the user needs to create the supporting files for a GitHub repository beyond the rule file itself and the README. This includes choosing and writing the LICENSE file, writing CONTRIBUTING.md with submission format and quality gates, creating .github/PULL_REQUEST_TEMPLATE.md, setting up the correct directory structure, and creating optional files like CHANGELOG.md and SECURITY.md. Trigger when the user says 'set up my repo files', 'what other files do I need', 'create a CONTRIBUTING file', 'scaffold the repo structure', 'set up the directory layout', or 'I have the skill and README, what's next'. Also trigger when the user is about to publish and asks what supporting files a proper repo should have. Do NOT trigger for writing the README itself (that is write-readme), for creating banner images (that is create-visual-assets), or for the git and GitHub operations (that is publish-to-github).

Create, revise, validate, package, or publish AI skills and rule files. Use as the main entry for the user's skill-authoring workflow, including creating a new SKILL.md, improving an existing rule, and preparing a skill repository. Choose only the requested stage; ordinary product code or content work is outside this skill.

Use this skill when the user wants to submit their published AI skill or rule repository to community directories and awesome-lists to increase visibility. This includes identifying which directories accept the repo's format, formatting the entry correctly per each directory's CONTRIBUTING.md, and walking through the fork → add entry → open PR process. Trigger when the user says 'submit to awesome list', 'how do I get my skill listed', 'add to the community directory', 'submit a PR to awesome-cursorrules', 'how do people find my skill', or 'promote my repo'. Also trigger right after the repo has been published and the user asks what to do next. Do NOT trigger before the repo is live on GitHub — the repo URL must exist before submitting anywhere. Do NOT trigger for the Claude Plugin Marketplace submission process, which requires a separate .claude-plugin/marketplace.json configuration.

Initialize Super Astra and select useful Astra guidance for ordinary task assignments, implementation, research, writing, questions, and follow-ups. Use when installing or first invoking the skill, when the host router supplies its catalog, or when a user asks to adjust agent behavior. Select modules by task intent before working; no complaint or keyword is required. Simple requests can use no modules. Do not turn ordinary tasks into configuration audits.

Use this skill when the user wants a bilingual or multilingual README for their AI skill or rule repository. This includes translating README.md into a second language as a separate file (README.zh-CN.md, README.ja.md), adding the language switcher line to the top of both files, and keeping the two versions in sync after edits. Trigger when the user says 'add a Chinese README', 'make my repo bilingual', 'translate the README', 'add 中文 docs', 'I want English and Chinese versions', or 'add a language toggle'. Also trigger when the repo targets a bilingual audience and the README exists in only one language. Do NOT trigger for translating the rule file or SKILL.md content itself (rule files stay in one language so agent behavior is predictable), or for writing the original README from scratch (that is write-readme), or when no README exists yet.

Write or renovate a repository README that explains its purpose, helps readers choose and use its tools, and matches its visual identity. Use for a new repository introduction or an existing landing-page improvement; preserve a narrow edit's scope.