Psychological Parent-Child Communication: Delegating Adolescent Psychological Assessment and Parent-Child Dialogue Strategies to Agents
This article introduces the "Psychological Parent-Child Communication" Skill, published by the enterprise account "Kunlun Growth" at the coordinates @org-zbk7xk96/edu-psych. Version 1.0.0, classified under education, it requires no API Key. Its core capabilities consist of three parts: adolescent psychology recognition, parent-child communication scripts, and conflict response frameworks. It operates through a progressive structure: "identify first, select scripts second, and switch to the response framework when conflicts escalate." The article outlines its positioning, functions, official acquisition address, and page URL. It honestly notes that the scraped materials lacked the SKILL.md body and specific installation commands; therefore, it did not fabricate integration steps for Cursor, Codex CLI, or Claude Code, but instead provided format references and suggested relying on the official page. It also cautions that this Skill is a general framework and does not constitute psychological counseling or diagnosis; situations involving self-harm, violence, or severe depression should be directed to professional help. Finally, noting the Skill's install count is 0 and its recent release, it advises users to test it personally and review the source files before implementation.
Read MoreMiaoshou ERP - TikTok Category Recommendation: Matching Leaf Categories and Required Attributes for Collection Cart Products
This article describes a Skill released by Shenzhen Chengyun on SkillHub. Coordinates: @org-ko8itrop/miaoshou-erp-tiktok-category-recommend. Version: 1.0.1. Source: enterprise. Category: business-ops. It is tagged with "requires_api_key". The page displays a score of 100000, 1077 downloads, and 0 installations. It recommends leaf categories and required attribute values for items in the collection box for the Miaoshou ERP TikTok Shop cross-border e-commerce platform. It is based on product titles, descriptions, detail IDs, sites, shop IDs, category trees, and category metadata. It triggers when users mention category matching, AI category recommendation, category attributes, recommended attribute values, TikTok categories, TK categories, category confidence, missing category attributes, or require category guidance before editing, writing back, or publishing. This article explains its positioning, core functions, installation and activation precautions, and typical usage.
Read MoreAI Video Generation for Architectural Walkthroughs: Full-Process AI Generation of Architectural Visualization Videos
Architectural Flythrough AI Video Generation is a community-sourced Agent Skill in the design-media category, located at @user_442551c3/architectural-flythrough-video-600029, version 1.0.0, published by user_442551c3. It compresses the production of architectural showcase videos into a single command: covering four types of videos including architectural exterior flyovers, interior space walkthroughs, landscape and garden showcases, and city skyline videos. It can be triggered via terms such as architectural flythrough, interior tour, and landscape showcase, with metadata indicating requires_api_key=false. This article provides trigger methods, three reproducible request formulations (exterior, interior, landscape), and notes limitations such as its community source, initial version, zero installs, and the official absence of published resolution and output format specifications. The official page currently fails to fetch; capability descriptions are based on pre-fetched materials, and installation commands should follow the official address instructions.
Read Moresdfadf: Kingsoft Docs Online Document Full Lifecycle Management Skill
"sdfadf" is a community Skill on SkillHub that consolidates six categories of Kingsoft Docs—documents, spreadsheets, presentations, multi-dimensional tables, collection forms, and attachments—into a unified invocation interface covering creation, editing, management, permissions, version history, and the recycle bin. It additionally supports format conversion, plain text extraction, push notifications, multi-dimensional table Webhooks, and batch asynchronous tasks. The Skill requires an API Key with the source listed as "community", is published by a personal account, and its page displays a rating of 100,000, 998 downloads, and 0 installations. Based on the official page description, this article does not fabricate installation commands or invocation examples, and notes that the publisher's claim of "covering all 8 major features of Tencent Docs" has not been independently verified.
Read Moredsh-prompt-clarifier: Rewrite prompts and clarify ambiguities before sending
dsh-prompt-clarifier is a DSH workflow plugin maintained by Duanyy0813 that adds a "Prompt Optimize" button to the input area. Before a message is sent to the assistant, it rewrites the prompt as a side-channel step based on the conversation context, and pops up clickable question cards at critical ambiguities. Progress is shown via tool cards without consuming model context, and if it fails, the original input continues as-is. It supports one-click installation via npm with dsh-prompt-clarifier@0.3.6, and is configurable for automatic mode, minimum word count, askBlocking, and more. Released under the MIT license.
Read MoreBuilder Exam Practice Bank · Professional Project Management and Practice: Skill for Generating Chapter-wise Exercises
SkillHub Community Skill @user_af163eaa/builder-quiz-m04-practice, version 1.0.1, categorized under education. This skill generates practice questions for the "Professional Engineering Management and Practice" exam for First/Second Class Constructors by chapter, covering construction techniques, project management, and practical regulations. It utilizes a universal framework that allows for specifying disciplines such as architecture, municipal engineering, or mechanical and electrical. The questions are AI-generated based on the general syllabus and are not exam predictions or official content. No API Key is required; after installation via coordinates on the client, questions are requested by chapter and professional direction. It is suitable for question-banking feature prototypes and personal exam preparation practice, though manual verification is required for formal scenarios.
Read Moredsh-claude-subscription: Running the Built-in Anthropic Routing with Claude Code Subscription in DSH
dsh-claude-subscription is a DSH bundle plugin maintained by dshapp. On macOS, it reads Claude Code (Pro/Max/Team) subscription credentials from the Claude CLI login keychain and passes them to the built-in Anthropic provider, eliminating the need for a separate Anthropic API key. The installation command is `dsh plugin --profile web add @dshapp/dsh-claude-subscription`; after installation, the Harness must be restarted. On first use, simply run `claude` locally to complete browser login. It supports selecting models via the Models page or declaring a model list through `cordis.patch.yml`. Tokens are renewed in the background and shared with the CLI. This plugin is intended for developers who have subscribed to Claude Code and run DSH locally. Before installation, please review the source code and MIT license, and be aware of the subscription terms and process permissions.
Read More[PWR] Perfect Employee Series · Operations: The Ideal Value of Full-Stack Operations Competency
Introduce the Community-sourced Operations Skill [PWR] Perfect Worker Series · Operations: Full-Stack Competency Ideal Values (located at @user_fa14b5bc/perfect-ops; version 1.0.1; category: business-ops; tag: requires_api_key=false). This Skill defines full-stack competency for operational roles using an ideal-value framework. It spans five capability domains—user, content, event, channel, and data operations—and maps a five-tier pyramid from Operations Specialist to COO. It applies across seven industries (Internet, manufacturing, retail, finance, healthcare, education, and public/enterprise sectors) and four company lifecycle stages from startup to decline. Built-in cross-domain coordination rules R1–R6 and exception-handling protocols E1–E3 are included, and all operational actions are driven by a user-lifecycle lens. The article outlines its positioning, core functions, official activation entry point, and framework-based usage, and notes that its outputs do not include specific budgets or real-time data; it serves as a framework reference rather than a directly executable operational plan.
Read MorePrimary School Essay Grading Skill: An Essay Grading Assistant for Teachers of Grades 3-6
"Elementary School Composition Review" is a Skill designed for teachers of grades 3 to 6, published by user_19f368e6 in the SkillHub community. It is categorized under "education," is currently at version 1.1.0, and requires no API Key. It supports three input methods: photos of compositions, pasted text, and readable documents. The output includes annotations with original image numbering, suggested grades, student feedback, and a teacher's reference. It also facilitates follow-up correction reviews, targeted mini-exercises, printable grading reports, and the management of student growth portfolios. The article introduces the core features, the method for enabling it by coordinates within the client, three reproducible typical use cases, as well as applicable scenarios and precautions: suggested grades require teacher review, and the growth portfolio involves student data that must comply with school management requirements. With a current installation count of 0, it is recommended to conduct a small-scale trial first. Official URL: https://api.skillhub.cn/user_19f368e6/elementary-composition-review
Read Moredsh-quicknote: Capture notes from Composer, manage on a single page
dsh-kit has released dsh-quicknote (v0.1.1, MIT), a DSH client plugin that captures drafts from the Composer toolbar as local notes. It allows users to search and filter Open/Done notes in a side panel, and provides agents with four tools, including quicknote_add. Data is stored in ~/.dsh/quicknote/notes.json. To install for the Web profile, run: dsh plugin --profile web add dsh-quicknote. Requires DSH >= 0.2.1-rc.5.
Read MoreShan Hai Jing (Outdoor Hiking AI Assistant): Empowering Agents to Leverage Hiking Knowledge Bases
This article introduces a community-maintained Agent Skill titled “Shan Hai Jing (Outdoor Hiking AI Assistant),” which organizes fifteen topic areas—hiking routes, gear purchasing, safety and first aid, trip leadership, physical training, nutrition and fueling, weather and date selection, transportation connections, navigation/GPS, outdoor photography, visas and permits, budget estimation, teammate recruitment and splitting expenses, descent/exit plans—into skill documentation. Trigger terms such as hiking, mountaineering, Yubeng, EBC, altitude sickness, hypothermia, GPX, and Tibet Travel Permit enable the Agent to answer hiking-related queries within its declared scope. The official materials provide coordinates, versions, sources, and categorizations, with no API key required. The article outlines the capability scope, installation and activation methods, typical questions, and caveats, and notes limitations including zero installs, no institutional endorsement, and the absence of real-time data sources.
Read MoreDSH Provider Hub: Unified Configuration of BharatCode, OpenAI-compatible Endpoints, and Subscription Models in Harness
DSH Provider Hub is a community plugin for DeepSeek Harness that centralizes configuration of BharatCode presets, OpenAI-compatible endpoints (including Chat Completions/Responses and local gateways) in the Settings→Providers panel. It supports integration with dsh-plugin-subscriptions to enable ChatGPT Codex and GitHub Copilot subscription authentication. The plugin uses the native llm-pi-ai adapter with Remote API credentials stored in DSH's credential vault. Installation requires enabling the @deepseek-ai/dsh-llm-pi-ai profile and executing "dsh plugin add github:arzvaak/dsh-provider-hub" following README instructions, with optional installation of dsh-plugin-subscriptions. Requires DSH version 0.2.0-rc.2 or higher.
Read MoreExcel PivotTables: Multi-Dimensional Aggregation of Detail Data
The "Excel Pivot" Skill released by Kunlun Growth (@kunlungrowth/excel-pivot, version 1.0.0) is categorized under office-efficiency and requires no API key. Targeting detailed data in Excel spreadsheets, it performs multidimensional summarization based on user-specified rows, columns, values, and filters. It outputs grouped data, percentage shares, and trend results, eliminating the need to rewrite aggregation scripts for every table. The article outlines core features, installation and activation methods, typical usage, applicable scenarios, limitations, and provides the official link.
Read Moredsh-deckforge: Creating Native PowerPoint Presentations in DeepSeek Harness
dsh-deckforge (Deckforge) is a workflow plugin maintained by DSH-Extensions-list. It provides integrated presentation capabilities for creation, refinement, and delivery within DeepSeek Harness. It compiles structured pages into native OOXML, keeps text and charts editable, and supports exports such as HTML and PDF. The toolchain covers from `deck_new` to `deck_export`, accompanied by G1–G6 quality gates and receipts. The sidebar Deck Studio shares the same revision model as the workbench. Installation requires cloning the repository followed by `dsh plugin install`; Node ≥22.19 is required. It is MIT licensed.
Read MorePost Race v2: A Source-Verifiable Community Skill
Post Race v2 is a community design-media Skill on SkillHub, located at @user_d6405f7b/race-v8-1787147998050284000 (Version 2.0.0, published by user_d6405f7b). This article exclusively presents facts verified from the scraped data, specifically the category (design-media), requires_api_key=false, the official URL, and the SkillHub page. As details regarding functionality, installation commands, configuration fields, and typical usage were not provided in the source material, no speculation is made. The article also notes discrepancies in ratings and download counts between leads and the scraped data, with the scraped data being treated as authoritative.
Read Moredsh-redact: sanitize credentials before tool outputs enter models and session logs
dsh-redact is a DeepSeek Harness plugin (admin-security) maintained by drm-nz that rewrites tool result text during the tools/post-execute stage, so that credentials enter the model and session logs in the form [REDACTED:NAME] without blocking commands or narrowing tool permissions. It supports four layers of rules: exact, patterns, context, and keywords, covering common cloud provider and API key formats as well as environment variables. The installation command is `dsh plugin --profile desktop add github:drm-nz/dsh-redact`, and configuration is written in `cordis.patch.yml`. The article clarifies the boundary: it applies only to tool output, does not revoke content already sent, and the source code and MIT license should be reviewed before installation.
Read MorePortfolio Skill: Generate Investment Plans Based on Funding Sources and Risk Tolerance
Released by a community developer, the **Portfolio Designer Skill** (ID: `@user_fc999673/touzizuhe`, version 1.0.0) is a professional-grade, API-key-free tool. Its core logic models two distinct capital scenarios separately: periodic contributions and lump-sum investments. The periodic contribution version requires monitoring safety guardrails only at the start and yearly thereafter; otherwise, it runs automatically in four equal parts. The lump-sum version prioritizes Money Market Funds (MMF) and short-term bonds for liquidity, then executes staggered purchases based on market signals including A-share ERP/PE, US VIX/Shiller PE, gold moving averages, and treasury yields. Any uninvested capital after 12 months is automatically deployed into a conservative portfolio. Long-term strategies utilize a Glide Path to gradually reduce equity exposure over time. The Skill follows a sequential six-step inquiry—covering funding source, goal, duration, risk tolerance, amount, and urgency—to generate an investment profile, recommended allocation weights, specific fund codes/fees, an execution plan, projected outcomes, and key considerations. Designed for direct conversation or slash-command triggers in **Claude Code**, it also supports **Codex CLI**, **Windsurf**, and **Tongyi Lingma** by loading `portfolio-designer.md` as Agent instructions. The developer disclaims that past performance does not guarantee future results and notes that this plan does not constitute investment advice.
Read MoreContent Creation Expert: From Topic Selection to Publishing Checklist, Generating Finished Content from a Single Scenario and Problem.
Content Creation Master (@user_ff29c4df/content-creation-master, v1.2.0, community-sourced, category: content-creation, no API key required) is an Agent Skill that standardizes the content creation workflow. It covers five stages—topic selection, copywriting, scriptwriting, formatting, and publishing—and produces titles, body text, scripts, and publishing checklists. It adapts content according to platform rules and annotates differences, explicitly avoids fabricating data, and does not involve professional editing software operations. The article first explains the pain points it addresses (difficulty in selecting topics, AI-generated-sounding copy, published articles that go unread), then covers its core functions, installation and activation details (including the official page and API address; since no installation command was provided in the pre-fetched materials and differences across hosting environments were not covered, none are fabricated here), and typical usage (can be triggered with a single sentence specifying platform, topic, and question, with follow-up questions for iterative revisions), and finally provides suitable use cases, four caveats, and the official URL.
Read MoreTeacher Bukeng's HTML Presentation: Delivering Presentations as Offline HTML Files
This introduction clarifies the positioning and usage of the Skill "bukenglaoshi-building-html-presentations" (The Teacher Who Doesn't Suck at Building HTML Presentations). Published by user_af3b8a3f (Version 1.0.0), it is sourced from the community and categorized under office-efficiency. It is identified by the coordinate @user_af3b8a3f/bukenglaoshi-building-html-presentations and requires no API Key. It solves the problem of delivering presentations as a standalone HTML file by enabling the Agent to generate a ready-to-use single-file presentation directly, eliminating the need to produce multiple HTML pages and stitch them together manually. Verified capabilities include multi-style support, offline single-file operation, compatibility with presentation remotes, and step-by-step narrative support. The article outlines suitable scenarios and notes specific limitations: the source is community-based rather than officially reviewed by the platform; with only thousands of downloads and zero installs, ecosystem validation remains insufficient; and it outputs HTML rather than editable PPT files. regarding installation, the text candidly states that since the source material does not include the SKILL.md body or installation commands, no unverified commands are provided. It only confirms that the Skill is hosted on SkillHub; after integration, it should be referenced by its coordinate and verified to appear in the available list. Finally, the official URL and the SkillHub page link are provided.
Read MoreSupplier Development Assistance: Creating a Closed Loop for Low-Performing Supplier Improvement Plans
Introducing the “Supplier Development Assistance” skill on SkillHub: a tool published by Gongchengke for generating support and improvement plans for low-performing suppliers. It produces documents in TXT and MD formats using a closed-loop of diagnosis-measures-tracking-verification, targeted at SQEs and supplier quality managers. Version 1.0.1, categorized as business-ops, no API key required. The article verifies the official introduction and metadata, and explains how to verify information not provided on the page, such as installation commands.
Read Moredsh-model-advanced: Set default inference strength and image handoff by model
dsh-model-advanced is a DSH model inference plugin maintained by drm-nz. It adds an Advanced card at the bottom of Settings → Models, allowing you to set default inference levels individually for each provider/model, and to designate a delegate for models that cannot read images. At runtime, it decorates paths such as resolveModelInfo and resolveCallConfig; images are transcribed once and written into a note with a handoff marker, supporting follow-up questions via inspect_image. Install it with the command dsh plugin --profile desktop add github:drm-nz/dsh-model-advanced (a restart is required); the configuration namespace is model-advanced.
Read MoreIn-depth Individual Stock Research Report Batch Generator: Research Report Generation Covering Multiple Stocks in a Single Pass
Introducing the "Deep Individual Stock Research Report Batch Generator" on SkillHub: Given a list of A-share, Hong Kong stock, or US stock codes or names, it batch generates and deeply refines research reports (in Markdown) for each stock, saving them as independent files. Capabilities cover westock-data acquisition, parallel sub-agent generation, a 13-chapter standard reporting framework (including three deep-dive directions: historical evolution and major events, industry chain positioning and business details, technical accumulation and future outlook), valuation, risk warnings, and data red lines. Source: community; Publisher: user_782d50e6; Version: 1.0.0; Category: professional; requires_api_key: false.
Read MoreMinimalist Paper-Texture Posters: Compile a Sentence or Photo into a 3:5 Paper Poster
The Minimal Paper Feeling Poster is a design-media category SKILL.md skill, located at @user_7f6c97a1/minimal-zine-poster, version 1.0.0. It transforms a single sentence, text, emotion, article, object, photo, or reference image into a default 3:5 vertical aged-paper-style poster with 70%-90% whitespace, a single visual focal point, a single high-saturation color accent, and a short caption. It also supports outputting only a four-part image generation prompt or reusing the visual system. It supports 5 modes: generation, photo input, reference image analysis, prompt-only, and generate-after-analysis. Draws inspiration and structural references from gc-minimal-zine-poster-v0-3 (MIT), has been localized to Chinese and adapted for WorkBuddy. The retrieved materials do not provide specific installation commands and configuration fields. It can be confirmed that requires_api_key=false.
Read Moredsh-novel-beat: Delegating Narrative Decision Points to the Author Before Drafting the Main Text
dsh-novel-beat is a DSH workflow plugin maintained by DrizzleGig. It lists decision points between scene outlines and main text based on the rule "ask whenever a story state change occurs." It accomplishes querying, persisting to disk, and compilation through five tools: beat_plan, beat_decide, beat_gate, etc. It optionally integrates with novel-forge. This article introduces the decision rules, workflow, data paths, and installation commands from the README, and explains the plugin's standalone operation and the necessity of reviewing source code before installation.
Read MoreError Analysis: A Skill for Converting Mistakes into Knowledge Gap Lists
Mistake Analysis (username: @kunlungrowth/edu-mistake-analysis, v1.0.1, Enterprise source, category: education, publisher: Kunlun Growth) is aimed at students, parents, and teachers. It diagnoses the causes of incorrect answers in homework and quizzes, pinpoints knowledge-point gaps, generates actionable, personalized remediation plans and variation drills, and outputs a knowledge-gap checklist for downstream skills such as learning-path planning and review outlines. The official tag requires_api_key=false indicates that no additional API key is required. This article describes its three-layer capabilities, capability boundaries, verifiable information, and a typical processing pipeline, and notes that the official page shows zero installs and lacks publicly available implementation cases; it recommends validating output quality with real mistaken problems before integration.
Read More