DockerとDocker Composeで開発環境を構成し、通信・データ保存・権限を見直すほか、元のファイルを変更せずにCLIの導入手順を検証するスキル。
- アプリとDBの開発環境を構築
- コンテナの通信とデータ保存の確認
- Docker設定のセキュリティ見直し
4,795 件 ・ 関連度順
概要と使いどころ
DockerとDocker Composeで開発環境を構成し、通信・データ保存・権限を見直すほか、元のファイルを変更せずにCLIの導入手順を検証するスキル。
Next.js + Better Auth + PostgreSQL を Docker で構築し、Cloud Run へデプロイするスキル。ローカル開発環境のセットアップから Docker Compose、Dockerfile 作成、Cloud Run + Cloud SQL + Secret Manager を使った本番デプロイまでをカバーする。「Next.js と Better Auth でアプリを作りたい」「Docker で PostgreSQL を使いたい」「Docker で構築したい」「Docker Compose を使いたい」「アプリを Docker 化したい」「Dockerfile を書きたい」「Cloud Run にデプロイしたい」「Cloud SQL や Secret Manager を使いたい」ときに使う。
Docker makes "works on my machine" a deployment strategy. Package your app with its dependencies, run it anywhere. But Docker's simplicity hides real complexity. A naive Dockerfile can be 10x larger and slower than it needs to be. This skill covers Dockerfile optimization, multi-stage builds, Docker Compose for development, security hardening, and production patterns. Key insight: your Dockerfile is code. Review it, optimize it, version it. 2025 lesson: Containers aren't VMs. The patterns that work for VMs (install everything, big base images) are anti-patterns for containers. Think small, think immutable, think layers. Use when "docker, dockerfile, container, docker compose, docker build, containerize, docker image, multi-stage build, docker network, docker volume, docker, containers, devops, deployment, infrastructure, security, optimization" mentioned.
日本語の概要は準備中です。原文の説明を表示しています。
Use when: User wants to extend Docker with custom tools, personalize the Docker environment, or set up user-specific Docker customization. Triggers: 'extend docker', 'docker-extend', 'add tools to docker', 'customize docker', 'add my tools to the container', 'personalize docker setup', 'docker user setup', 'install tools in docker'. Does: Interactively sets up Dockerfile.user and docker-compose.override.yml so users can add personal tools to their Docker environment without affecting maintainer files or committing user-specific config to git.
日本語の概要は準備中です。原文の説明を表示しています。
Train and run models on Tinker (Thinking Machines) — LoRA and full fine-tuning, SFT with correct loss masking, RL/RFT with group-relative advantages and importance sampling, agentic RL over multi-turn tool-using agents, inference via the SamplingClient or the OpenAI-compatible endpoint, checkpoint management and export, every hyperparameter, and cost estimation from the live price table. Use this skill whenever the user mentions Tinker, tinker://, TINKER_API_KEY, console.tinker.ai, or the tinker CLI; wants to fine-tune, SFT, RFT, or RL a model on Tinker; asks about Tinker models, context windows, LoRA rank, sampling params, loss functions (cross_entropy, importance_sampling, ppo, cispo, dro), Datum construction, forward_backward or optim_step; wants to sample from, download, resume, merge, or export a Tinker checkpoint; or asks what a Tinker run will cost. Whenever Tinker work is initialized in a project, this skill also creates a TINKER_PRICING.md there so prices sit next to the training code. Reach for it even on vague asks like "train a model on this data" or "get inference working" when Tinker is the platform in play.
日本語の概要は準備中です。原文の説明を表示しています。
🇺🇸 Product Launch & Product Hunt #1 Playbook — Multi-channel launch timing and sequencing. T-14 warm-up → T-0 launch day → T+7 momentum. Hunter selection, maker comment templates, comment cadence, anti-shadowban rules. Distilled from 30+ Product Hunt #1 daily launches across 150+ AI startups. **Supports 3 product paths: developer/maker (PH+GitHub), B2C consumer (Reddit/communities/TikTok, skip PH), B2B (LinkedIn+media).** 🇨🇳 产品发布与 Product Hunt 夺冠手册 — 多渠道发布的时序与节奏。T-14 预热 → T-0 发布日 → T+7 后续。Hunter 选择、maker comment 模板、评论节奏、防 shadowban。提炼自 30+ 次 Product Hunt 日冠、150+ AI 创业公司实战。**支持三种产品路径:开发者/maker(PH+GitHub)、2C 消费者(社区种子/TikTok,跳过 PH)、B2B(LinkedIn+媒体)。** 🇯🇵 プロダクトローンチ&Product Hunt 1位プレイブック — マルチチャネルのローンチ時系列。T-14準備→T-0当日→T+7モメンタム。ハンター選定、メーカーコメント、コメントリズム、シャドウバン回避。**3つの製品パス対応:開発者/maker(PH+GitHub)、2C消費者(コミュニティ種・TikTok、PHスキップ)、B2B(LinkedIn+メディア)。** 🇰🇷 제품 출시 & Product Hunt 1위 플레이북 — 멀티채널 출시 타이밍과 시퀀싱. T-14 워밍업 → T-0 출시일 → T+7 모멘텀. 헌터 선정, 메이커 코멘트, 댓글 리듬, 섀도밴 방지. **3가지 제품 경로 지원: 개발자/maker(PH+GitHub), 2C 소비자(커뮤니티 씨드/TikTok, PH 건너뜀), B2B(LinkedIn+미디어).** Triggers: "product launch" | "launch strategy" | "Product Hunt" | "PH launch" | "how to launch" | "go to market" | "GTM" | "launch day" | "find a hunter" | "maker comment" | "launch sequencing" | "发布策略" | "产品发布" | "发 PH" | "Product Hunt 准备" | "找 hunter" | "ローンチ" | "출시 전략"
Use when: User wants to extend Docker with custom tools, personalize the Docker environment, or set up user-specific Docker customization. Triggers: 'extend docker', 'docker-extend', 'add tools to docker', 'customize docker', 'add my tools to the container', 'personalize docker setup', 'docker user setup', 'install tools in docker'. Does: Interactively sets up Dockerfile.user and docker-compose.override.yml so users can add personal tools to their Docker environment without affecting maintainer files or committing user-specific config to git.
日本語の概要は準備中です。原文の説明を表示しています。
OpenClawのDocker環境で一連の動作を確かめるテストを作成・変更するスキル。模擬サーバーと実サービスを使い分け、送信内容や最終応答まで検証します。
Set up worker environment and per-worker state for parallel pools. Use when code needs paths, environment variables, database connections, loaded libraries, or expensive objects available on workers before parfor/parfeval runs. Teaches parallel.pool.Constant, parfevalOnAll, and parpool name-value pairs. Also use when refactoring existing code that uses spmd for side-effect setup (an anti-pattern). Triggers: worker setup, pool constant, per-worker state, non-serializable, loadlibrary on workers, database connection parfor, addpath workers, spmd before parfor, worker environment, reduce parfor overhead, parfor setup, resource creation in parallel loop, cannot serialize error, undefined function or variable on workers error, load data per worker, reduce data transfer, parallelize setup, improve parallel code.
日本語の概要は準備中です。原文の説明を表示しています。
临床诊断思维 (临床诊断思维 / 临床推理 (Clinical Diagnostic Reasoning) — 医生『怎么想病』的元学科:从症状/体征/检查到诊断结论的认知操作系统,从业者(临床医生/住院医/规培生,尤其全科/急诊/大内科等以未分化主诉为生的科室)、医学生与医学教育者、诊断安全与质量改进研究者、以及做医疗 AI 辅助诊断产品的人的视角。覆盖: (a) 第一性张力 — **直觉模式识别 (System 1: illness scripts 疾病脚本 / pattern recognition / gestalt, 『资深人一眼认出 aunt Minnie』) ⇄ 分析性推理 (System 2: hypothetico-deductive 假设演绎 / Bayesian 概率更新)** 的 dual-process 双过程理论 (Kahneman→Croskerry/Norman 谱系), 资深≠更会分析而是『脚本库更大+校准更好』; 更深层论战 — **『认知去偏可教 (Croskerry: bias awareness / cognitive forcing strategies / diagnostic timeout) ⇄ 偏倚标签是马后炮、知识结构才是主因 (Norman/Sherbino/Monteiro: debiasing 干预 transfer 证据弱, bias 是 hindsight 标签)』** — 本行最核心学术对垒; 『概率思维 (验前概率 × 似然比 → 验后概率, Pauker-Kassirer test/treatment threshold 阈值模型) ⇄ 穷尽式排查 (rule-out everything / 防御性医疗 / VOMIT)』; 『临床 gestalt ⇄ 结构化临床决策规则 (Wells/PERC/HEART)』; 『诊断简约 Occam's razor ⇄ Hickam's dictum (病人可以同时得 N 个病)』; 『床旁体格检查复兴 (Verghese Stanford 25 / McGee 循证体检) ⇄ 影像检验替代床旁』; (b) 方法论正典 — illness script theory (Schmidt/Boshuizen: enabling conditions/fault/consequences 三段结构), problem representation + semantic qualifiers 问题表征与语义限定词 (Bowen NEJM 2006, 把病人翻成 one-liner), hypothetico-deductive model (Elstein 1978《Medical Problem Solving》: 早期假设生成+定向检验), Bayesian 工具箱 (sensitivity/specificity/LR/Fagan nomogram/SnNout-SpPin), threshold model (Pauker-Kassirer NEJM 1975/1980), 认知偏倚分类学 (anchoring/premature closure/availability/confirmation/base-rate neglect/search satisficing/diagnostic momentum — Croskerry 偏倚清单), 去偏与元认知策略 (diagnostic timeout/cognitive forcing/Ely checklist/calibration), schema-based reasoning (Clinical Problem Solvers schemas: 按 pivot 症状走分支), reflective practice 结构化反思 (Mamede/Schmidt), 诊断错误科学 (NAM 2015 报告定义 / Newman-Toker Big Three / Hardeep Singh e-triggers / SAFER Dx 框架); (c) 行业结构与角色 — 医学生→实习/住院医 (晨会 morning report/查房被 pimping/汇报训练)→主治 attending→master clinician (NEJM CPC discussant/晨会大师如 Dhaliwal); 配套生态: 医学教育者 (clinical reasoning curriculum + 评估: script concordance test/key features exam/OSCE), 诊断安全研究者 (SIDM/AHRQ), 诊断辅助与 CDS 工具开发者 (DDx generator/AI); (d) 核心工作流 — 数据采集 (病史为王 + 循证体检 + 针对性检验影像) → 问题表征 (one-liner + semantic qualifiers) → 鉴别诊断生成 (schema / 解剖定位法 / VINDICATE-M 病因筛, 按『常见可能 × 致命不能漏 can't-miss』双轴排序) → 假设定向检验 (按 LR 选检查 / 阈值决策) → working diagnosis + 安全网 (red flags 交代 / test of time / test of treatment) → 反馈校准 (follow-up / M&M / diagnostic timeout); 教学工作流: 晨会渐进披露汇报 / CPC / SNAPPS / one-minute preceptor / virtual morning report; AI 增强工作流 (2023-2026: LLM 鉴别诊断头脑风暴 / OpenEvidence 检索 / ambient scribe 释放认知带宽 + 自动化偏倚 guardrails); (e) 产出物 — one-liner, problem list, prioritized DDx, assessment & plan (按问题分层), 晨会/CPC 汇报, M&M 复盘, 诊断不确定性沟通 (『最可能是 X, 但出现 Y 红旗立刻回来』); (f) 教育与评估 — script concordance test / key feature exam / EPA / 里程碑; 中国语境: 人卫《诊断学》教材 / 执业医师考试 / 规培结业临床思维考核; (g) 争议/批判 — debiasing 之争 (Croskerry vs Norman-Sherbino『knowledge is the cure』, 去偏 RCT 效果弱 / bias 标签不可证伪), 诊断错误率数字之争 (『10-15%』经典估计 / Newman-Toker 79.5 万美国年严重伤害外推方法被质疑 / 尸检符合率), dual-process 二分被批过度简化 (连续谱 / 难以实证分离), 决策规则 vs gestalt (资深 gestalt 常不输 Wells 类规则 / 算法厌恶), vignette 研究外推性 + context/case specificity (推理高度内容绑定不可通用迁移 [Norman/Eva] — 对『教推理通用课』产业的根本批判), AI 辅助诊断 2023-2026 (GPT-4 在 NEJM CPC/vignette 追平或超医生 [Kanjee 2023 JAMA / Goh 2024 JAMA Netw Open『AI alone > physician+AI』悖论] / Google AMIE / OpenEvidence 爆发 vs 自动化偏倚 / 去技能化 / 真实环境验证缺失), 防御性医疗与过度检查 (incidentaloma 瀑布 / VOMIT), 体检衰亡之争, pimping 教学法争议; (h) 流派/思想谱系 — 决策科学/Bayesian 派 (Ledley-Lusted 1959 → Elstein/Kassirer/Sox/Pauker/Eddy → Brush) vs 认知心理/双过程派 (Kahneman-Tversky → Croskerry/Graber) vs 教育认知派 (Schmidt/Boshuizen illness scripts → Norman/Eva/Mamede/Durning 情境认知) vs 诊断安全/系统派 (NAM 2015 / Newman-Toker / Hardeep Singh / Graber-SIDM: 错误=系统×认知共因) vs 床旁临床派 (Osler 传统 → Tierney aphorisms / Verghese / McGee / Dhaliwal / Saint: 病史体检为王 + 刻意练习) vs AI/计算派 (INTERNIST-1/DXplain 专家系统 → Isabel → LLM 世代 Rodman/Topol)。不含: 具体专科治疗方案与手术决策深度 (足踝外科/种植牙等另有 skill), 中医辨证 (另有 skill), EBM 文献批判性评价方法学 (相关但本 skill 聚焦诊断推理本身), 护理诊断体系 (NANDA), 影像/病理知觉型读片训练 (相邻但独立), 精神科 DSM 结构化访谈细节, 医患沟通技巧全集 (只覆盖诊断不确定性沟通), 患者自查/自我诊断指
日本語の概要は準備中です。原文の説明を表示しています。
德州扑克策略 (德州扑克策略 (Texas Hold'em) — solver/GTO 时代的无限注德州扑克竞技打法,牌手/职业视角。覆盖: (a) 理论两极 — GTO 博弈论均衡 (不可剥削的基线策略) vs Exploitative 剥削打法 (针对对手漏洞偏离均衡最大化 EV),以及二者的实战取舍; (b) solver 时代工作流 — PioSOLVER/GTO Wizard 跑解、范围构建、节点锁定 (node-locking) 做剥削、用 HUD+数据库 (Hold'em Manager/PokerTracker/Hand2Note) 复盘找漏洞 (leak finding); (c) 牌型分层 — 现金局 (cash game) vs 锦标赛 (MTT) 的根本差异,后者叠加 ICM (独立筹码模型) 与跳台/泡沫期博弈; (d) 核心概念栈 — 范围 (range)/权益 (equity)/EV/位置 (position)/翻前开池与 3-bet/4-bet/翻后下注理论 (c-bet、极化 vs 浓缩范围、MDF 最小防守频率、阻断牌 blockers、超池下注 overbet)、SPR、底池赔率; (e) 元层 — 资金管理 (bankroll management)、桌位选择 (table selection)、方差与心态 (mental game/tilt 控制)、抽水 (rake) 与场次选择对长期赢率 (winrate, bb/100) 的影响; (f) 智识演进 — 从 Sklansky 古典手牌价值 → Janda/Tipton 博弈论 → solver 普及 → GTO Wizard 民主化求解 → AI 超人 (Libratus/Pluribus) 对人类策略的反哺。学派分歧: GTO 派 vs 剥削派、现金局 vs 锦标赛、理论自上而下 vs 牌局自下而上、solver 纯学院派 vs 实战感觉派。不含: 线下牌场运营/发牌荷官、博彩合规与赌场管理、扑克之外的牌类 (桥牌/斗地主)、纯概率赌博 (老虎机/轮盘)。) Master OS — automated mastery of 德州扑克策略 (Texas Hold'em) — solver/GTO 时代的无限注德州扑克竞技打法,牌手/职业视角。覆盖: (a) 理论两极 — GTO 博弈论均衡 (不可剥削的基线策略) vs Exploitative 剥削打法 (针对对手漏洞偏离均衡最大化 EV),以及二者的实战取舍; (b) solver 时代工作流 — PioSOLVER/GTO Wizard 跑解、范围构建、节点锁定 (node-locking) 做剥削、用 HUD+数据库 (Hold'em Manager/PokerTracker/Hand2Note) 复盘找漏洞 (leak finding); (c) 牌型分层 — 现金局 (cash game) vs 锦标赛 (MTT) 的根本差异,后者叠加 ICM (独立筹码模型) 与跳台/泡沫期博弈; (d) 核心概念栈 — 范围 (range)/权益 (equity)/EV/位置 (position)/翻前开池与 3-bet/4-bet/翻后下注理论 (c-bet、极化 vs 浓缩范围、MDF 最小防守频率、阻断牌 blockers、超池下注 overbet)、SPR、底池赔率; (e) 元层 — 资金管理 (bankroll management)、桌位选择 (table selection)、方差与心态 (mental game/tilt 控制)、抽水 (rake) 与场次选择对长期赢率 (winrate, bb/100) 的影响; (f) 智识演进 — 从 Sklansky 古典手牌价值 → Janda/Tipton 博弈论 → solver 普及 → GTO Wizard 民主化求解 → AI 超人 (Libratus/Pluribus) 对人类策略的反哺。学派分歧: GTO 派 vs 剥削派、现金局 vs 锦标赛、理论自上而下 vs 牌局自下而上、solver 纯学院派 vs 实战感觉派。不含: 线下牌场运营/发牌荷官、博彩合规与赌场管理、扑克之外的牌类 (桥牌/斗地主)、纯概率赌博 (老虎机/轮盘)。: top builders' mental models, tool stack, current workflows, jargon, and where to keep up. Trigger this skill when the user works on 德州扑克策略 (Texas Hold'em) — solver/GTO 时代的无限注德州扑克竞技打法,牌手/职业视角。覆盖: (a) 理论两极 — GTO 博弈论均衡 (不可剥削的基线策略) vs Exploitative 剥削打法 (针对对手漏洞偏离均衡最大化 EV),以及二者的实战取舍; (b) solver 时代工作流 — PioSOLVER/GTO Wizard 跑解、范围构建、节点锁定 (node-locking) 做剥削、用 HUD+数据库 (Hold'em Manager/PokerTracker/Hand2Note) 复盘找漏洞 (leak finding); (c) 牌型分层 — 现金局 (cash game) vs 锦标赛 (MTT) 的根本差异,后者叠加 ICM (独立筹码模型) 与跳台/泡沫期博弈; (d) 核心概念栈 — 范围 (range)/权益 (equity)/EV/位置 (position)/翻前开池与 3-bet/4-bet/翻后下注理论 (c-bet、极化 vs 浓缩范围、MDF 最小防守频率、阻断牌 blockers、超池下注 overbet)、SPR、底池赔率; (e) 元层 — 资金管理 (bankroll management)、桌位选择 (table selection)、方差与心态 (mental game/tilt 控制)、抽水 (rake) 与场次选择对长期赢率 (winrate, bb/100) 的影响; (f) 智识演进 — 从 Sklansky 古典手牌价值 → Janda/Tipton 博弈论 → solver 普及 → GTO Wizard 民主化求解 → AI 超人 (Libratus/Pluribus) 对人类策略的反哺。学派分歧: GTO 派 vs 剥削派、现金局 vs 锦标赛、理论自上而下 vs 牌局自下而上、solver 纯学院派 vs 实战感觉派。不含: 线下牌场运营/发牌荷官、博彩合规与赌场管理、扑克之外的牌类 (桥牌/斗地主)、纯概率赌博 (老虎机/轮盘)。 problems and wants industry-grade thinking, tool selection, or workflow guidance. 触发词:「德州扑克」「德扑」「Texas Hold'em」「poker strategy」「扑克策略」
日本語の概要は準備中です。原文の説明を表示しています。
Cloud Workerで使うツールを事前に組み込んだイメージを作成・選択し、起動時間と稼働を検証。失敗時の切り戻しや、承認後の旧イメージ削除も扱います。
Webアプリの本番公開に向けて、Docker構成やテスト・公開の自動化を整え、更新方式の選定から稼働確認、障害時の復旧手順まで支援するスキル。
Webアプリの本番公開に向け、CI/CDによるテストと配布の自動化、Docker化、稼働確認、切り戻し手順、公開前の確認項目を整理するスキル。
Best practices for Docker-based ROS2 development including multi-stage Dockerfiles, docker-compose for multi-container robotic systems, DDS discovery across containers, GPU passthrough for perception, and dev-vs-deploy container patterns. Use this skill when containerizing ROS2 workspaces, setting up docker-compose for robot software stacks, debugging DDS communication between containers, configuring NVIDIA Container Toolkit for GPU workloads, forwarding X11/Wayland for rviz2 and GUI tools, or managing USB device passthrough for cameras and serial devices. Trigger whenever the user mentions Docker with ROS2, docker-compose for robots, Dockerfile for colcon workspaces, container networking for DDS, devcontainer for ROS2, multi-stage builds for ROS2, or deploying ROS2 in containers. Also trigger for CI/CD with Docker-based ROS2 builds, CycloneDDS or FastDDS configuration in containers, or shared memory in Docker. Covers Humble, Jazzy, and Rolling distributions across Ubuntu 22.04 and 24.04 base images.
日本語の概要は準備中です。原文の説明を表示しています。
Container specialist for Docker, Docker Compose, image optimization, and container orchestration fundamentalsUse when "docker, dockerfile, container, docker-compose, image, containerize, docker build, multi-stage build, docker, containers, dockerfile, docker-compose, images, kubernetes, devops, containerization, microservices" mentioned.
日本語の概要は準備中です。原文の説明を表示しています。
World-class container image building - Dockerfiles, multi-stage builds, security hardening, and the battle scars from images that broke in productionUse when "docker, dockerfile, container, image, docker-compose, build, multi-stage, alpine, distroless, scratch, docker build, docker run, registry, ecr, gcr, dockerhub, layer, cache, docker, containers, dockerfile, images, containerization, devops, cloud-native, microservices" mentioned.
日本語の概要は準備中です。原文の説明を表示しています。
Write production-grade Dockerfiles, docker-compose configurations, multi-stage builds, and container optimization. Activate on "Dockerfile", "docker", "docker-compose", "container", "multi-stage build", "docker image", "container optimization", "docker security". NOT for Kubernetes orchestration, cloud-specific container services (ECS, Cloud Run), or CI/CD pipelines (use github-actions-pipeline-builder).
日本語の概要は準備中です。原文の説明を表示しています。
claude-task-worker のカスタムワーカー(`workerFiles` に登録する TS 定義)を、`AskUserQuestion` で要件を全項目確定させてから生成し、`claude-task-worker list-workers` でロード検証までするスキル。「カスタムワーカーを作って」「独自のワーカーを追加したい」「新しいラベルで動くワーカーを定義したい」といったリクエストで使用する。
Write production-grade Dockerfiles, docker-compose configurations, multi-stage builds, and container optimization. Activate on "Dockerfile", "docker", "docker-compose", "container", "multi-stage build", "docker image", "container optimization", "docker security". NOT for Kubernetes orchestration, cloud-specific container services (ECS, Cloud Run), or CI/CD pipelines (use github-actions-pipeline-builder).
日本語の概要は準備中です。原文の説明を表示しています。
NotionのページをMarkdownで取得・作成・更新し、データの絞り込みや検索を行います。ファイル、コメント、Workersの操作も扱うスキルです。
Imports a pedestrian into CARLA on UE 5.8 as a spawnable, animating walker — imports the skinned FBX bound to CARLA's shared pedestrian skeleton, duplicates a donor walker blueprint and repoints it at the new mesh, and registers it in WalkerParameters.json as walker.pedestrian.<id>. Can also export a shipped walker to FBX, which is how you obtain a rig-conforming mesh to start from. Use when the user asks to "import a walker/pedestrian", "add a custom character", "clone a walker", or has a walker that spawns but stands still.
日本語の概要は準備中です。原文の説明を表示しています。
Provides guidance for writing and benchmarking optimized CUDA kernels for NVIDIA GPUs (H100, A100, T4) targeting HuggingFace diffusers and transformers libraries. Kernels must be kernel-builder/ABI3-compliant: no pybind11, no setup.py, TORCH_LIBRARY_EXPAND bindings only. Supports models like LTX-Video, Stable Diffusion, LLaMA, Mistral, and Qwen. Includes integration with HuggingFace Kernels Hub (get_kernel) for loading pre-compiled kernels. Includes benchmarking scripts to compare kernel performance against baseline implementations.
日本語の概要は準備中です。原文の説明を表示しています。
ALWAYS activate when the user's query involves Docker in any way — even if it also matches other skills. If the words docker, Dockerfile, docker-compose, compose.yml, container, or image appear in the query, this skill MUST be used. Covers: writing or editing Dockerfiles and compose files, adding services (postgres, redis, etc.) to compose, volume mounts and data persistence, docker build failures (layer caching, npm install issues), healthchecks and service startup ordering (depends_on), environment variables in containers, port mapping, container crashes and exit codes (OOM/137), non-root users, multi-stage builds, image optimization, .dockerignore, and deploying to container runtimes. Takes priority over general implementation or debugging skills when Docker infrastructure is the subject.
日本語の概要は準備中です。原文の説明を表示しています。