ninehills

Reviews code diffs and release-ready changes after implementation, runs on-demand project-wide code-quality audits with a 4-axis scorecard, executes approved implementation plans, extracts project-specific constraints from repository context, auto-fixes safe issues, and drives approved release, publish, push, release-reaction, and issue/PR follow-through. Also triages issues and PRs when the user mentions them. Not for exploring ideas, debugging, or document prose review.

Produces distinctive, production-grade UI for components, pages, and visual interfaces, with artifact-grounded screenshot iteration for real app polish. Not for backend logic or data pipelines.

Generate visually striking PPT slides via OpenAI's gpt-image-2 -- 10 curated styles (Spatial Glass / Tech Blue / Editorial Mono / Dark Aurora / Risograph / Wabi / Swiss Grid / Hand Sketch / Y2K Chrome / Retro Vector) plus a template-clone mode that mimics any user-supplied .pptx; ships an HTML viewer and a 16:9 .pptx. Use when the user asks to make a presentation, slides, deck, pitch deck, investor PPT, magazine-style PPT, or 做一份 PPT / 生成幻灯片 / 用 gpt-image 生成 PPT / 按这个模板生成 PPT.

Runs a budget-aware Agent Health audit for Codex, Claude Code, agent instructions, verifier surfaces, and AI maintainability when agents ignore instructions, hooks/MCP fail, validation is missing, or AI-written code is hard to maintain. Flags issues by severity. Not for debugging code or reviewing PRs.

Finds root cause of errors, crashes, regressions, screenshot-reported defects, unexpected behavior, and failing tests before applying any fix. Not for code review or new features.

Runs a six-phase research workflow to turn unfamiliar domains or collected sources into publish-ready output. Not for quick lookups or single-file reads.

Fetches any URL or PDF as clean Markdown for reading, quoting, citation, or downstream work. Handles paywalls, JS-heavy pages, X/Twitter, and Chinese platforms via proxy cascade. Not for local text files already in the repo.

Turns rough ideas into approved, decision-complete plans with validated structure before writing code. Covers new features, architecture decisions, and value judgments about whether to build, keep, or remove something. Not for bug fixes or small edits.

生图 / 生成图片 / 画图 — 用 OpenAI gpt-image-2 生成图像。支持文生图、参考图生图 (img2img)、蒙版修补 (inpainting)。当用户要求用 GPT 画图、OpenAI 生图、gpt-image-2、文+图生图、参考图片生成、img2img、inpainting 时必加载此技能。Auth 自动继承 OPENAI_API_KEY / Codex OAuth (Pi/Codex) / .env / config.yaml。

Remove AI writing patterns from English prose. Covers blog posts, technical docs, social copy, release notes, scientific writing. Trigger: deslop, de-AI, make it sound human, remove AI patterns, fix slop, polish, rewrite, sound natural

去除中文写作中的 AI 腔调，让文本像人写的而非模型生成的。覆盖博客、技术文档、推文、发行说明、科学论文等场景。触发词：去 AI 味、改写、润色、审稿、翻译腔重、像人写的、自然一点。英文去 AI 味见 deslop-en

Use when reviewing code changes, PRs, diffs, or commits. 触发：审查、review、CR、看看这段代码、检查这个 PR、审一下。

为任意 coding agent 编写高质量任务目标的指南。当用户提到「设定目标」「goal」「任务目标」「持久目标」「完成标准」、或需要为复杂多轮任务（性能优化、调试、迁移、研究复现等）定义可验证的完成条件和约束时触发。

Control herdr from inside it. Manage workspaces and tabs, split panes, run commands, read output, and wait for state changes — all via CLI commands that talk to the running herdr instance over a local unix socket. Use when running inside herdr (HERDR_ENV=1).

Jupyter Notebook 创建与格式转换技能。当用户提到以下任何请求时必须使用：新建/创建 notebook、Jupyter notebook、.ipynb、jupytext、py:percent、# %% cell 格式、notebook 版本控制、paired notebook、notebook 转 Python、Python 转 notebook、用纯文本写 notebook。即使用户只说「新建一个 notebook」也应触发此技能——先询问用户选用 .py percent format 还是传统 .ipynb，再按选定格式生成。不用于调试已有 notebook 或写普通 Python 脚本。

当用户明确要求头脑风暴、收集需求、写 PRD、或探索「要不要做 / 做什么」时使用。通过协作对话将粗略想法转化为清晰的产品需求文档。不自动触发——仅在用户显式发起需求探索流程时才启动。

将 PPT/PPTX/PDF 演示文档转换为拟人化逐字稿（Markdown + JSON 双格式输出），并将逐字稿回写到 PPTX 演讲者备注中。使用五阶段流水线：素材提取、大纲生成、逐字稿撰写（通过 Gemini CLI）、合并输出、备注回写。当用户需要将演示文档转为演讲稿、朗读稿、解说词时使用此技能。触发短语包括但不限于：「PPT 转语音」「生成逐字稿」「演讲稿生成」「帮我把 PPT 做成配音稿」「生成演示文档的解说词」「presentation to speech script」「PPT 转文字稿」「从 PPT 生成演讲词」。即使用户没有明确提到逐字稿，只要意图是从演示文档生成可朗读的文字稿，也应触发此技能。

将 PPT 分页转换为 PNG 图片和 Markdown 文字稿（包含演讲者备注）。使用场景：处理 PPT 演示文稿，提取每页内容和演讲者备注为独立文件。

生成符合中国专利法规范的专利交底书。接收 Markdown 格式的技术描述作为输入，通过多阶段流水线（标题生成、专利检索、发明目的撰写、发明内容撰写、图表生成、报告合并）自动生成完整的专利交底书。

根据技术交底书、DOCX/PDF/Markdown 技术材料，生成完整的中国专利申请文件草稿，包括摘要、权利要求书、说明书、附图说明和具体实施方式。

Use when the task asks for a visually strong landing page, website, app, prototype, demo, or game UI. This skill enforces restrained composition, image-led hierarchy, cohesive content structure, and tasteful motion while avoiding generic cards, weak branding, and UI clutter.