Coding & Development

Turn a pasted list of rough feature ideas into well-formed tracker issues, with a per-item debate before anything is filed. Use when the user hands over a batch of ideas, a backlog dump, a list of fixes, or says "file these as issues", "make tickets from this", "turn this list into issues", "batch these features", "create a backlog from this", "I have a list of things to do". Debates each item (restate, ask at most two questions with a proposed default, pin acceptance criteria), shows one review table for the whole batch, and only then files issues with labels. GitLab and GitHub.

Enforce Rule Cascade rules in Go services and tools (net/http, chi, gin, echo, gRPC, workers) with rulescascade.com/go. Use when wiring rules into a Go code path, replacing validate tags or hand-written checks with the engine, or mapping rule findings to HTTP or gRPC errors.

Enforce Rule Cascade rules on the JVM - Java 17+, Kotlin and Scala; Spring Boot, Quarkus, Micronaut, Jakarta EE - with com.rulescascade:rules-cascade-core. Use when wiring rules into a JVM service, replacing Bean Validation or hand-written checks with the engine, or mapping rule findings to HTTP errors.

适用于 AI 多 Agent 与人类团队协同研发的多角色状态流转、质量审计、防错闭环与看板自动化工作流技能包。可通过 /fy-flow help、/fy-flow start、/fy-flow status、/fy-flow kanban、/fy-flow trace 快捷指令或自然语言唤醒。

为国产麒麟 Linux 与奇安信可信浏览器进行前端兼容开发、排错和评审。涉及麒麟、银河麒麟、中标麒麟、奇安信、信创、Qianxin、Kylin、旧 Chromium，或在已声明此目标环境的项目中新增/修改页面、CSS、Tailwind、组件库、JS/TS、构建配置、浏览器 API、字体、弹窗、下载与打印时使用；覆盖源码和构建产物检查、能力降级及真机验证。

On-demand dual-model review via native `spawn_subagent` (defaulting to mcli models `deepseek-v4.1-flash` and `glm-5.3`, or user-specified models) for standalone or combined research actions (Plan, Coding, Test, or Coding+Test together). Use when designing an experiment plan, reviewing post-training/RL/masking code changes, or auditing test validity.

Use when a task has already been completed and accepted once and the user wants to repeat it or solidify the process — recurring deliverables, weekly reports, multi-source research, or a workflow that currently depends on one person remembering the steps.

Reconstruct flat images, screenshots, posters, diagrams, or infographics as editable PowerPoint slides while preserving the source composition and visual effect. Use when the user asks to convert an image into an editable PPT/PPTX, split image elements into editable layers, recreate a screenshot in PowerPoint, translate in-image text into editable slide copy, or says 图片转可编辑PPT、把图片元素拆出来、保留当前效果、图片重绘成PPT、截图转PPT.

Set up and maintain a product lens — one git-tracked JSON graph of a product's purpose, core loop, avatars, interviewees, evidence, claims (Heard / seen · Owner's view · Untested), focus hypotheses with computed confidence, shipped surfaces and ideas, alternatives and competitors, market insights, content pillars and decisions — built into one self-contained, commentable index.html with a diff view. Use when asked to create a lens, add an interview or other evidence, update a surface after changing it, add an idea from a brainstorm, add a competitor, or show what changed in the lens.

Migrate an nf-core pipeline from the vendored utils subworkflows (utils_nextflow_pipeline, utils_nfcore_pipeline, utils_nfschema_plugin) and local helper copies to the nf-core-utils Nextflow plugin, or bump nf-core-utils / nf-schema to a newer version. Use this whenever someone wants to "use nf-core-utils", drop or replace vendored nf-core utils subworkflows, move helpers like getGenomeAttribute, softwareVersionsToYAML, paramsSummaryMultiqc, completionEmail or methodsDescriptionText to `plugin/nf-core-utils`, update the nf-core-utils or nf-schema pin, or build MultiQC tool citations from module meta.yml with citationsOnTheFly (publication blocks), or debug a run failing with "Missing process or function X ... Did you mean 'X'" after a plugin change (including AWS megatests / Seqera Platform runs).

Use when interacting with ANSYS AEDT through the current PyAEDT CLI. Triggers on requests to start or inspect AEDT sessions, manage projects, create designs, run scripts, export screenshots or configurations, or use the PyAEDT command line from the workspace virtual environment.

Perform an on-demand, bounded documentation review and update while keeping canonical docs accurate and lean.

用 PyAEDT 驱动 Ansys HFSS / AEDT 完成天线与 RF 仿真自动化的工作流骨架。当用户提到 HFSS、Ansys Electronics Desktop、AEDT、pyaedt、ansys.aedt.core,或要做天线仿真、S 参数(S11/S21)、回波损耗(return loss)、VSWR、远场方向图(far-field)、增益、辐射效率、阻抗匹配、参数化扫频(frequency sweep)、参扫优化(parametric sweep),或者说"自动化跑 HFSS / 用脚本驱动 HFSS / 让 agent 帮我跑仿真"时,务必使用本 skill。即使用户没明说"用 pyaedt",只要任务核心是"用 Python 控制 HFSS 做天线/RF 仿真",也应触发。

Automate ANSYS HFSS tasks on Windows through PyAEDT. Use when Codex needs to inspect the local HFSS environment, create or modify HFSS projects, set design variables, build simple geometry, create setups and sweeps, run analyses, or export Touchstone results from `.aedt` projects.

Entry point for work on the Nextora block-theme ecosystem — child themes, Gutenberg blocks, design presets and the visitor colour/font switcher, and template/part/pattern composition. Use when a request mentions Nextora or a Nextora child theme, when the task spans more than one of those areas, or when you need the source map and production quality gates for this ecosystem before writing code. Also use to decide which specialist skill a multi-part request needs.

Create, bootstrap or audit a child theme of the Nextora parent block theme — the style.css header with Template nextora, the functions.php load order, the single version constant kept in sync with the header, the front-end and editor-canvas asset pipeline with its parent handle dependencies, the child block registrar and inserter category, PHP/i18n conventions, and the optional feature-file layout. Use when a request names a Nextora child, a new child theme name or slug, a child theme header or version bump, child CSS/JS enqueueing, editor canvas styles, a child block folder or block category, porting Nextsaas conventions into a new child, or asks why child code cannot call the parent at file scope. Pair with nextora-design-system for any child that must follow palette or visitor font/colour switching.

Prepare the current branch for review and open a GitHub pull request: run /code-review and fix its findings, review and fix the docs, run /security-review and fix its findings, record the branch's design decisions (ADRs) as commit trailers, then push and open (or update) the PR with the gh CLI, following the repo's PR template if it has one. Stops and reports, without pushing, if a finding can't be fixed. Use when the user says "/pr", "review and PR this branch", or "review, fix and open a PR". Accepts an optional base branch, a review effort (low|medium|high|xhigh), and --draft.

Full review round over every open pull request of a GitHub repository in one command. Reviews each PR on its own with a full report of what changed and the issues found, detects conflicts between PRs (who conflicts with whom, in which files and lines), predicts which PRs will conflict after a merge, and lets the user accept and merge, reject, or comment on each PR. Use when the user asks to review all open PRs, run the PR round, check PR conflicts before merging, or invokes /pr-round.

Crea o pone al día el informe de una tarea deliberativa de la Software Factory (investigar, deliberar, diseñar) — con la estructura fija de resumen, el análisis, lo decidido, lo que queda por decidir partido en bloques, las separaciones propuestas y las tareas que salen — y con las preguntas como fichas con opciones, recomendación destacada y estado (abierta, decidida, separada), al día con cada respuesta del propietario. Cada vuelta sube la versión, respalda la anterior en _respaldo/ y regenera el HTML. Úsala cuando una tarea deliberativa escriba o actualice su informe, cuando haya que registrar en él una respuesta, una decisión o una separación, o al pasar un informe antiguo a este formato. Usa el generador y los recursos comunes de crear_documentos.

Crea o revisa el documento fundacional de un proyecto o de un producto — su presentación formal, más formal que el documento de visión y sin historia conductora — con la necesidad documentada, la tesis de desarrollo, qué es el producto, los escenarios de uso, las evidencias con lo que respaldan y lo que no, qué debe demostrar el piloto y las fuentes. Para el autor y para compartir con colaboradores, socios o financiadores. Úsala cuando pidan un documento fundacional, una presentación o justificación formal del proyecto, un manifiesto, un dossier o un caso de desarrollo, o convertir uno existente (PDF, HTML) a este formato. Genera .md → .html con el generador y los recursos comunes de crear_documentos (portada sobria, secciones plegables, tarjetas, fichas con buscador, referencias) y el tema de la identidad del proyecto; se imprime en A4.

文档问答助手工厂——把任何产品手册/操作规程/制度文档装载为可追溯的专属Q&A助手，仓库自带一份虚构的《桌面气泡水机》演示知识库。回答带逐字原文编号出处（M001–MNNN）可回查；插图/视频能作为文件直接发就优先发文件，发不了才报章节位置，绝不描述或OCR图片内容；网址/路径/参数等具体值必须能在原文找到出处，找不到就明说文档没写，绝不编造。换装新文档：提供 DOCX 跑一条 calibrate 命令即可自动重标定（逐字原文编号、图片按章节归位、媒体索引重建、引用改写全自动）。当用户要求"根据这份文档建问答助手/答这份手册里的问题/把文档变成专属助手"，或询问当前已装载手册（演示：桌面气泡水机）的安装、制作、滤芯更换、E1/E2 故障、童锁等内容时使用。

Ciclo de vida de las tareas de un proyecto bajo la Software Factory (SF) y su cola, .sf/peticiones.json — estados (en cola, en curso, en observación, descartada), pasos, ramas feature/NN-nombre, subagentes, prueba y cierre por squash en develop, descarte, reversión, reapertura, eliminación, fusión y separación, historial por tarea y copias de la cola — con el CLI tools/cola.py. Úsala siempre que haya que tocar la cola — anotar una petición o un bug, poner una tarea en curso o dejarla en cola, crear su rama, añadir o marcar pasos, lanzar o recoger un subagente, dejarla lista para probar, cerrarla, pasarla a observación o devolverla a ella, cerrarla o revertirla en observación, descartarla, reabrirla, eliminarla, fusionar o separar tareas, o recuperar una cola rota.

Load the minimum sufficient context from a repository-owned doc_llm before planning, implementation, review, or debugging. Use when the user names an Epic/Subtask, asks to initialize context, or the active repository already uses DLM. Do not borrow state from another repository or trigger for repositories without doc_llm unless explicitly requested.

Listen with earshot whenever you must judge, choose, trim, verify or compare sound you cannot hear - picking a sound effect, music or ambience from candidates, finding the cleanest moment in a recording, checking whether audio reads as a real recording or as code-synthesized, reviewing a soundtrack or a video's audio (what happens when, sync to picture, loudness, mix), A/B-ing a mix change, making a sound in code and improving it by listening (earshot forge), or giving a code-made film its sound effects from the shipped bank (earshot sfx / kit). Use it before telling anyone how audio sounds.