codex-skills

Route CEO-first multi-window work into owner-led role loops. Use for 总控/CEO、架构/CTO、内容主编、角色派发与复用、模型路由、来源窗口回调、role-windows 台账、skill 命中统计或 multi-window loop engineering。

Audit and optimize a skill system with evidence. Use for skill architecture, token or context waste, role and prompt boundaries, routing hits or misfires, plugin packaging, registry/docs/source drift, duplicate workflows, or deciding whether a reusable change belongs in a skill. Runs deterministic checks first and applies only authorized, minimal fixes.

用打分、盲预测、发布后复盘和 rubric 进化校准内容判断。适用于视频、文章、播客等可量化内容；触发词包括初始化、打分、预测、发布、复盘、推荐选题、抓热点、状态、找对标。首次使用先运行 cheat-init。

为最终稿创建不可变盲预测日志，并在已有预测时追加新版本。通过 cheat-score-blind 做隔离评分，再由主窗口复核落盘。触发词包括启动预测、start prediction、打分并预测、写预测日志。

提议并执行 rubric 或 bucket 升级。两种模式：**完整 rubric bump**（最高风险动作，5 步强制 + 跨模型审核）和 **--bucket-only 轻量重校**（只换 bucket 边界，不动 rubric 公式）。**Phase 2 强制走 cheat-score-blind sub-agent 给校准池重打分**——不接受 self-scored fallback。触发词："升级 rubric"/"bump rubric"/"更新公式"/"我想加一个维度"/"调整权重"/"重校桶"/"recalibrate bucket"。

cheat-on-content 的首次 onboarding 与脚手架创建器。统一流程——所有用户都走相同 5 阶段闭环，唯一区别是"发过视频的人"会在 init 时多一步：抓取已有视频建立历史 context（用于后续 cheat-seed 给更贴合的选题、更准的 baseline）。触发词："初始化"/"init"/"首次使用"/"我是新用户"/"setup cheat-on-content"。**必须在用户第一次会话执行；其他子 skill 在 .cheat-state.json 不存在时自动路由到此。**

从对标账号导入 script + 数据 → 拆 pattern + 派生 base rubric 信号 → 写到 benchmark.md / script_patterns.md / rubric_notes.md。**这是工具最早期信号的来源**——cold-start 用户没自己历史时全靠对标，发过历史的用户也建议至少 1 个对标做 sanity check。触发词："学这个账号"/"拆这几个对标视频"/"learn from"/"导入对标账号"/"找对标"。

把老用户的 .cheat-state.json 升级到当前 schema_version。读 migrations/registry.md 算迁移链，按顺序应用每一步迁移文件。幂等：跑两次结果一样。失败停在中间版本不前进。触发词："迁移"/"升级 state"/"migrate"/"我的 state 是老版本"/"schema 版本不对"。

登记一篇内容已发布，把 URL/平台 ID/发布时间写入对应预测文件 header 和 state file。这是一个轻量动作——只更新元数据，**不动预测段任何字符**。触发词："已发布"/"I shipped"/"发布链接是 X"/"刚发完 [url]"/"publish registered"。

从 candidates.md 里按当前 rubric 排序推荐 top N 选题，每条带 composite + 一句 rationale + 锚点对比。**candidates 不存在时给引导而非报错**。触发词："推荐选题"/"next topic"/"下一篇做什么"/"recommend topics"/"挑一个选题"。

T+N 天数据回收 + 复盘 + 把实绩观察写入 rubric-memo.md。这是校准循环的反馈环节——不复盘的预测等于占星。触发词："复盘 [path]"/"retro this"/"T+3d 数据来了"/"抓数据 [path]"/"把这篇复盘了"。

INTERNAL sub-agent, not a user-facing skill. Performs context-isolated nine-dimension rubric scoring for cheat-score, cheat-predict, and cheat-bump. Accepts only script and rubric-note paths, returns strict JSON, and refuses post-publish evidence.

给单篇稿子打 rubric 分。**只在控制台输出，不写文件，不预测**。触发词："打分这篇 [path]"/"score this [path]"/"给这稿子打分"/"先打分看看"。是 cheat-predict 之前的轻量探索动作。

登记一条视频已拍摄。**建 video folder + 询问实际拍摄稿是否与 scripts/<id>.md 一致 + buffer +1**。与 cheat-publish 配对：拍了进队列，发了出队列。触发词："拍了"/"拍了 X"/"shot"/"shot it"/"已拍 X"/"录完了"。

cheat-on-content 的状态看板。显示当前模式 / rubric 版本 / 校准进度 / 待复盘 / pool 状态 / 是否该升级 SQLite / 是否该 bump rubric。**任何时候都可调，无副作用**。触发词："状态"/"看板"/"status"/"我现在该做什么"/"进度怎么样"。

从配置的热点源（HN / Reddit / YouTube trending / B 站热门 / 等）抓今天的热门话题，去重 + 粗打分 + 写入 candidates.md。**绝大部分人没有候选池——这是让"我没素材"问题在 onboarding 第二步就消失的钥匙**。触发词："抓热点"/"fetch trends"/"今天有什么可做的"/"trending now"/"找选题"。

Use when splitting content work between a raw drafting model/window, an editor-in-chief, and the user's final taste review. It keeps prompts loose enough for judgment while preserving facts, account boundaries, and publish gates.

Use when a user edits or criticizes generated copy and you need to extract durable writing-style rules instead of merely remembering surface words.

Use when drafting, reviewing, or revising Xiaohongshu/Douyin/social-platform Chinese text so it has real internet-native judgment instead of README tone, technical-briefing tone, or mechanical meme-word stuffing.

Use when editing Chinese web fiction, narrative prose, or story drafts that sound AI-generated, overly polished, formulaic, explanatory, or too symmetrical; use to detect AI-writing fingerprints and revise only wording while preserving plot, facts, characters, names, and intent.

Operate the WeChat Official Account AI-application workflow for articles, weekly digests, rich formatting, draft-box updates, and local handoff. Use for Agents, MCP, Skills, tool calling, workflows, and open-source AI updates; exclude Xiaohongshu and Xianyu.

将Markdown文章转换为美化的HTML格式，适配微信公众号发布。应用专业CSS样式、代码高亮、优化排版。当用户说"美化这篇文章"、"转换为HTML"、"优化公众号格式"、"生成公众号HTML"时使用。

自动搜索、抓取、改写技术内容，生成适合微信公众号的中文科普文章。涵盖AI大模型、GitHub开源工具、技术话题。当用户说"写一篇关于XXX的文章"、"帮我研究XXX"、"生成公众号文章"时使用。

Run authorized Xiaohongshu browser workflows for login checks, creator preview fill, publish blockers, final posting, search, export, or approved engagement. Prefer native signed-in Chrome; keep Python/CDP for deterministic batch fallback.

Read and analyze Xiaohongshu comments from the user's signed-in Chrome session. Use to collect, export, summarize, or classify comments for content planning, follow-up posts, audience research, reply strategy, and 引流文案。

Prepare copy-ready Xiaohongshu/Rednote publish materials from a local XHS article package. Use when the user asks for "输出发布格式", "小红书发布格式", "准备发布", "二次编辑发布格式", or needs title/body/tags/output image directory/checks without browser automation or clicking publish.

Choose the narrowest reliable browser surface across Codex Browser, signed-in Chrome, Playwright CLI, and platform scripts. Use for browser automation, logged-in workflows, UI inspection, screenshots, form filling, publish previews, or frontend QA.

Pressure-test startup ideas against real user pain, current behavior, competitors, first-customer access, MVP scope, founder-market fit, and a two-week launch test. Produces a direct strong, weak, or pivot verdict.

GStack autoplan method for running a broad CEO, design, engineering, and developer-experience review before opening implementation windows.

GStack canary method for staged release thinking, smoke gates, rollout checks, and rollback signals before or after deployment.

GStack careful mode for risky work that needs a conservative pass, hazards, safeguards, and explicit confirmation points before execution.

GStack CSO method for broad infrastructure-first security posture review, threat areas, severity calibration, and safe remediation scope.

GStack design consultation method for critiquing product/UI/visual direction, exploring alternatives, and recommending a design path before production.

GStack design-html method for producing or planning quick HTML prototypes, design boards, or rendered visual explorations for UI review.

GStack design-review method for reviewing rendered UI, landing pages, decks, or visual artifacts for hierarchy, layout, accessibility, and polish.

GStack design-shotgun method for rapidly exploring multiple distinct design directions before choosing one to implement or polish.

GStack developer-experience review for implemented repositories and workflows, focused on setup friction, command reliability, docs gaps, and maintainability.

GStack document generation method for turning implementation, product, or operational work into durable docs with clear audience and ownership.

GStack release documentation method for changelog, release notes, announcement copy, and internal/external release communication.

GStack freeze method for pausing writes and limiting work to evidence gathering when a task is unsafe, ambiguous, or production-sensitive.