Developmental Screening and Special Needs Assessment for Children: Developmental Evaluation Skills for Early Childhood Educators and Parents
"Child Developmental Screening and Special Needs Assessment" is a developmental screening skill designed for kindergarten teachers, childcare providers, pediatricians, and parents. Coordinated at @user_4c5a13be/developmental-screening-assessment (Version 1.0.0), it is published under the education category by SkillHub community publisher user_4c5a13be. Based on authoritative tools such as the Child Developmental Scale, Gesell Developmental Schedules, and the Hong Xiumin Social Development Assessment System, it performs comprehensive developmental evaluations across five major domains: gross motor, fine motor, language, social behavior, and adaptive ability. It identifies tendencies for developmental delays or abnormalities and provides early intervention recommendations and support strategies. The tag indicates requires_api_key=false. Trigger scenarios include developmental screening, identifying developmental delays or abnormalities, assessing social development, and seeking early intervention advice. Since the documentation does not provide installation commands or reproducible examples, this article focuses solely on verified capabilities and identifiers, emphasizing that it is a screening tool rather than a clinical diagnostic instrument.
Read MorePresentation Consultant: Breaking Down PPT Creation from Requirements to Final Product into 8 Steps
The PPT Creation Advisor is an orchestrated, eight-module Agent Skill for PPT creation, located at @user_4f12e6bd/ppt-get. Classified under office-efficiency (Version 1.2.0), it requires no API Key. The main workflow is fixed to Requirement Understanding, Content Refinement, Outline Structure, Layout & Color Scheme, Graphics & Charts, Template Adaptation, Review & Quality Control (QC), and Export & Delivery. It incorporates three decision branch points (J1/J2/J3), a closed-loop QC feedback mechanism, and resumable execution capabilities. Each stage generates JSON intermediate artifacts to facilitate traceability and handoffs. It integrates all capabilities of the ppt-generator-skill, including 17 color palettes, an 18-style library, big data visualization, matrix layouts, and web search supplementation. It can auto-generate content when assets are missing and auto-complete information with cited sources when data is insufficient. The final output is an editable PPTX file.
Read MorePersonal Legal Assistant: Breaking Down Legal Issues into Verifiable Conclusions
Personal Legal Assistant is a Skill in the professional category on SkillHub, version 1.4.8, sourced from enterprise. It targets ordinary individuals encountering legal issues and lawyers needing to locate legal authorities. After the user explains the situation, it provides tailored analysis and solutions; it first identifies similar judgments and compensation ranges; when litigation is necessary, it can draft legal documents such as statements of claim and defenses or review and edit existing filings; it also searches statutes and regulations, offers clause-by-clause interpretation, and compares Chinese and foreign provisions. Its core capability is that every conclusion is accompanied by source links that can be traced back on a clause-by-clause basis, which is the main difference from generic conversational legal Q&A. Configuration requires an API key (requires_api_key=true). No one-click install command is provided; add it by following the platform’s instructions. It provides analysis and document-drafting support rather than legal services, and outcomes for individual cases are not guaranteed. For sensitive matters, information should be de-identified beforehand.
Read MoreAI Mentor for the Entire MEM Thesis Process: Breaking Down a Part-Time Master's Thesis into 28 Steps
The "MEM Thesis Full-Process AI Tutor" is a thesis coaching Skill designed for working graduate students (MEM and management-related professional master’s students). Published by SkillHub community author user_de867193, the current version is 2.1.1. Classified under education and tagged with `requires_api_key=false`, it requires no additional API Key application. It breaks down the entire master's thesis journey—from topic selection, literature review, research design, and data analysis to main text writing, plagiarism checking/similarity reduction, and defense slides—into 28 steps and 26 stage cards. Each step details the required outputs, methodologies, and self-assessment criteria, accompanied by checklists. The tool provides executable inspection tools for writing, plagiarism checking, formatting, data verification, and version control. It covers 24 industry backgrounds, including software, manufacturing, finance, healthcare, construction, government affairs, and education, offering specific data sources, metrics, charting techniques, and common defense questions for each sector. Supporting multi-platform environments such as OpenClaw, Codex, and WorkBuddy, the tutor enables users to bypass the need for preliminary research methodology training. Users simply provide their current status and materials; the AI identifies the current bottleneck and guides them through the subsequent steps. This article verifies the official profile, version, source, and rating/download data, while outlining the target audience and relevant limitations.
Read Moredsh-memory-tree: Transforming DSH Context Compression into a Hierarchical Memory Tree
The DSH plugin `dsh-memory-tree`, maintained by bgx666, transforms session compression from flat summaries into a layered memory tree. It features node merging, checkpoint root nodes, memory cards, and the ability to drill down into the source text via `explore_memory_tree`. It supports per-session toggles and parameter adjustments on the settings page. The installation target is `github:bgx666/dsh-memory-tree`, requiring DSH >=0.2.1-rc.3 and <0.3.0. This plugin is still under development; please refer to the README and license before use.
Read Moredsh-memory-hub: A DSH Local-First Cross-Session Memory Plugin
dsh-memory-hub is a DSH memory plugin released by the maintainer beijingwahw. It features local JSONL storage, automatic capture of subscribed sessions/events and tools/results, and provides four tools: memory_store, memory_recall, memory_forget, and memory_status. Based on the official README, this article explains how to install from a tgz package, configure profiles, and use importSources for data import. It also cautions that the plugin runs with the same permissions as dsh and recommends reviewing the source code before installation.
Read MoreAutonomous Browser Operation Agent: Automation Skill for Web Browsing, Data Extraction, and Batch Form Filling
This article introduces a community skill called "Autonomous Browser Operation Agent" (@user_c5d41a7d/browser-auto-agent, version 1.0.0, categorized as AI agent). It codifies the process of simulating human browser operations into SKILL.md, allowing the model to autonomously orchestrate the steps based on task objectives. Capabilities include web browsing, data extraction, bulk form filling, content publishing, page screenshots and screen recording, and asset downloading, supplemented by four mechanisms: human-like behavior simulation, session persistence, intelligent element locating, and automatic retry on task failure. During the writing process, both the SkillHub page and API failed to be retrieved. Therefore, this article only records verified locating information and trigger scenarios, deliberately omits fabricated installation commands and configuration fields, and recommends validating it on a small-scale task before deployment. The target audience includes operations, content, and data extraction personnel. Key considerations include the lack of institutional endorsement for community sources, its initial release status, zero installation count, compliance boundaries for data collection and bulk publishing, and the risk of locating failures caused by site redesigns.
Read MorePreparation for Fanwei Project Kick-off Meeting: Generating a Page-by-Page Confirmed Outline from Project Facts
A community Skill designed for the Weaver project kickoff meeting, located at @user_040dcc29/weaver-project-kickoff. It is categorized under business-ops, sourced from the community, and published by user_040dcc29 (Version 1.0.10). The documentation indicates that no API key is required. The workflow first prepares business facts item by item, then generates a page-by-page outline consistent with the final HTML and PPT text. After the user confirms the outline page by page, a standard rendering handover package is delivered. Both the HTML ZIP and PPT display upload options individually; files are only handed over to the document capability item by item—following a fixed reporting material directory—if the user explicitly selects to upload. Project planning, Logo handling, rendering, and uploading are executed by independent capabilities, ensuring the Skill does not overstep its boundaries. The documentation does not provide installation commands or explain loading differences for Cursor, Codex CLI, or Claude Code, requiring independent verification. The listed metrics show a rating of 100,000 and 1,185 downloads, but 0 installations. Zero installations imply a lack of actual usage feedback, meaning functional performance must be verified independently.
Read Moredsh-style-seed: DSH Session-Level "we need" Thinking Style Toggle
dsh-style-seed is an experimental DSH web plugin maintained by bartholo4, providing a "weneed" switch for each new session. When enabled, it temporarily injects authorized external dialogue examples into the initial request (including necessary tool continuations) to guide the model toward more frequent task-centric English thinking expressions, without mandating the use of "We need" at the beginning. It is compatible with DSH 0.2.2-alpha.2 and Node >=24.9.0, and is enabled and configured via `seedPath` in `cordis.patch.yml`. This is an experimental version with no guarantee of effectiveness; the service must be stopped and tasks must be completed before maintaining the plugin. Installation is performed via `dsh plugin add` after cloning.
Read Moreoneone-Market Research: Expanding the Agent's Retrieval Scope from the Chat Window to the Entire Web
oneone-Market Research (Agent Reach) is a general-purpose Agent Skill published by @user_48324e52 on the SkillHub community. Currently at version 1.0.3, it is categorized under data-analysis with the tag requires_api_key=false. It encapsulates the search capabilities of 17 platforms into a directory structure consisting of SKILL.md and references/*.md, covering six categories: search, social (Xiaohongshu/Douyin/Weibo/Twitter/Bilibili/V2EX/Reddit), career (LinkedIn), dev (GitHub), web (webpages/articles/WeChat Official Accounts/RSS), and video (YouTube/Bilibili/Podcasts). It provides four invocation entry points—CLI, MCP, curl, and Python scripts—with zero-configuration support for 8 channels. Built-in trigger words include market research, research, competitive analysis, search, and platform names; once identified by the Agent, it executes according to a routing table. This article introduces its positioning, capability layers, official coordinates, and activation methods, while pointing out limitations such as zero installations for version 1.0.3, the need to refer to categorized documentation for deep interactions with various platforms, and the fact that this Skill only addresses data acquisition rather than analysis.
Read MoreOffice Productivity Toolkit: A Collection of Deterministic Algorithms for Document Processing
Office Efficiency Toolbox (@user_053d27d0/office-helper-zx) is a SKILL.md format Skill published by SkillHub community developer user_053d27d0. It is version 2.0.0, categorized under office-efficiency, and tagged with requires_api_key=false. The official description claims it contains 10 advanced office algorithms; the page lists 7 specific items: TF-IDF file classification, PDF structure parsing, Myers diff algorithm, TextRank summarization, interval tree scheduling conflict detection, AVL balanced tree, and intelligent merging. The names of the remaining algorithms could not be verified. The article is structured in the following order: Preface, Overview, Core Functions, Installation & Activation, Typical Usage, Applicable Scenarios & Caveats, and Conclusion. It explains that the tool delegates tasks such as document classification, structure extraction, summarization, conflict checking, and version comparison to deterministic algorithms rather than model inference. During the writing process, both official URLs failed to load; therefore, installation commands and invocation examples are not provided. The article also truthfully notes the gaps in available information and limitations.
Read Moredsh-ask-first: The Composer clarification button that asks before rewriting
dsh-ask-first is a DSH client plugin maintained by BAIYE0303 that provides a "Clarify First" button in the bottom-left corner of the Composer: before sending, it restates its understanding, clarifies requirements through clickable questions, then rewrites the input into a request containing goals and completion criteria and fills it back into the input box, never sending on your behalf. To install, run `dsh plugin --profile web add dsh-ask-first` and restart the host. It requires no configuration or API Key, uses the model currently selected in the Composer, and a complete flow involves approximately two model calls. It is suitable for scenarios where requirements are still vague or lengthy typing is inconvenient; it can be skipped when you can already write clear requests or want the agent to start working directly.
Read Morelawyerai: Converting '2024 Lawyer's Guide to Using AI' into a Retrievable Legal AI Skill for Answering Questions
lawyerai (@user_fffe0f19/lawyerai, v1.0.0, community source, professional category) is an AI Q&A skill tailored for legal scenarios, encapsulating the full text of the "AI Guide for Lawyers 2024" as its retrieval corpus. When lawyers, legal counsel, or legal practitioners inquire about using tools such as Kimi, ERNIE Bot, Tongyi/Tongyi Farui, MetaLaw, MiaoTa, FaTianShi, ProcessOn, Xmind, ChatGPT, Zhipu, iFLYTEK, Doubao, Pkulaw, or Wolters Kluwer for legal research, document drafting, contract review, case law retrieval, similar case recommendation, client consultation, case analysis, due diligence, visualization, or office efficiency, or seek advice on AI management norms for law firms and data security compliance, the skill retrieves operational guidelines and scenario examples from the original text of the guide. The tag `requires_api_key=false` indicates that no separate API Key configuration is required. Current public materials do not provide specific installation commands for each client; the guide's scale (approximately 190,000 words, over 20 types of tools, and nearly 100 scenarios) is sourced from the Skill introduction. It is recommended to refer to the original book for accurate details before implementation.
Read MoreMarket Monitoring Co-pilot: Introduction to the Intelligent Personal Market Monitoring Skill for A-Shares and Hong Kong Stocks
Introducing the Market Monitoring Copilot Skill: an intelligent copilot for personal A-share and Hong Kong stock monitoring that is information-driven, human-in-the-loop, non-quantitative, and strictly avoids automated order execution. It covers a five-layer architecture, JSON schemas for holdings and funds, data source selection, 8 automation task templates, a local market monitoring workbench, report structures, and practical lessons learned from real-world pitfalls.
Read MoreHome Assistant Office Assistant: Organizing Repetitive Office Tasks into Structured Results
Home Assistant Office Assistant is an Agent Skill released by Zhiyu Technology, categorized under office-efficiency. It is based on the universal SKILL.md format, currently at version 1.0.0, with coordinates @org-02qudk26/home-assistant. Targeting white-collar workers, it is positioned to handle repetitive tasks related to office efficiency, producing structured and reusable work deliverables. The tag requires_api_key=false indicates that no additional API Key configuration is required for integration. Official page snapshots show a rating of 100000, 1264 downloads, and 0 installations, indicating very few actual users. The functional scope does not disclose a specific list of sub-features. It is recommended to confirm task compatibility and conduct a small-scale trial before integration. This article is based solely on verified official materials and does not provide unverified installation commands.
Read Moredsh-long-memory: DSH Automatic Long-Term Memory
The DSH plugin dsh-long-memory (v0.9.1) by baixiaoyin666 monitors completed conversation turns on the Host side, automatically extracts up to 3 global/workspace facts and writes them to local JSON. When a new session is initiated, it injects a summary of ≤1400 characters during the initial system prompt assembly. It features memory_search/save/forget functions, a details panel, diary, and persona. Installation involves adding a local directory and enabling it in settings, with no specific configuration options required. Requires DSH 0.2.x, licensed under MIT.
Read Moredsh-shot-ocr: Hotkey screenshot, local OCR fill into Harness input box
dsh-shot-ocr is a DSH client plugin maintained by baibaiyi, targeting Windows 10/11. It allows you to select a desktop region via global or in-app hotkeys, recognizes text locally using Windows.Media.Ocr and writes it into the Harness input box, or skips OCR to directly insert a screenshot attachment. It covers installation commands, three types of entry points, two output modes, screenshot region selection and the settings card, and addresses PowerShell 5.1, language packs, and common issues. OCR does not upload images; they are only sent into the conversation after you explicitly send them.
Read MoreLiki Professional Numerology Skill: Integrating Charting and Divination into Agents
The Liki Professional Fortune-Telling Skill (@user_357240c7/liki, v7.1.0, community, professional) is a community skill in the general SKILL.md format, designed to transform fortune-telling Q&A from free model generation to engine-based chart casting, rule-based judgments, and traceable rationale. Its capabilities cover BaZi fortune-telling, birth chart analysis, Four Pillars of Destiny, Zi Wei Dou Shu, major life cycles and annual fortunes, marriage, career, wealth, health, academics, and family relationships; Liu Yao divination,卦-based questioning, and timing analysis; Qi Men Dun Jia for strategic analysis and negotiation timing; almanac date selection (marriage, business opening, moving, house entry, renovation, travel); home and office feng shui, store location selection, Eight Mansions, and Xuan Kong Flying Stars; as well as baby naming, adult name changes, company naming, brand naming, and name testing. The tag requires_api_key=false indicates no additional API key is needed. According to the official page at the time of writing, it shows a rating of 100,000, 1,264 downloads, and 0 installations, which differs from earlier clues of 98,690 and 1,110—the page should be considered authoritative. The collected materials do not include specific installation commands or platform-specific instructions, and the article does not provide unverified commands; please refer to the official address page. The content is from a traditional cultural perspective and is for reference only, not constituting professional advice.
Read Morehuahua-dsh-memory-os: Adding Long-term Memory and Collaborative Orchestration to DSH Sessions
huahua-dsh-memory-os (HOS / dsh-memory-evolve) is a DSH plugin oriented towards memory management, maintained by azure5100. It offers features such as five-track long-term memory, to-do and project logs, Git-based synchronization for memories, multiple session rooms, dispatch to external agents via COI, prompt injection, bookmark branching, and session reviews. Based on the official README and package.json, this article outlines the environment requirements (DSH≥0.2.0, Node≥20), the official installation command, and typical conversational prompts. It also reminds users that the plugin runs with the same permissions as the DSH process and that the source code and MIT License should be reviewed before installation.
Read MoreAgnes Image Recognition: Enhancing Text-Only Agents with Visual Capabilities
Agnes Image Recognition (@user_d550b199/ack26814) is a community skill in the AI-agent category on SkillHub. By leveraging the Agnes AI multimodal model agnes-2.5-flash to parse images, it converts visual content into structured Markdown text, providing image understanding capabilities for text-only LLM agents. It supports local images, URL images, and multi-image comparison, covering OCR text extraction, chart interpretation, and UI screenshot analysis. The skill is compatible with CLI, pipes, and custom Python agents (supporting --stdin, --json, --quiet, and system prompts). API key management features three-level protection: environment variables, configuration files, and command-line arguments. It defaults to accessing the China site (api.agnes-ai.cn), supports the international site with automatic fallback, and has zero third-party dependencies.
Read MoreNatural Language Generation of Mini Program Pages: Transforming a One-Sentence Description into Complete WeChat Mini Program Page Code
This article introduces the community Skill "Generate Mini Program Pages from Natural Language" (@user_92725383/miniapp-page-builder, v1.0.0, category dev-programming). It accepts natural language descriptions and outputs the complete quartet of WeChat Mini Program page files: WXML structure, WXSS styles, JS logic, and JSON page configuration, with no API key required before use. The article explains its positioning and capability boundaries, provides checkpoints for installation and activation, a reproducible workflow from describing requirements to saving the page files, as well as applicable scenarios and limitations: generated results must be manually reviewed for event bindings, method names, and class names; complex business and security logic must be implemented by the user.
Read Moredsh-session-fork-tree: Organizing DSH sidebar forked sessions into a parent-child tree
This article introduces the `dsh-session-fork-tree` plugin by `ayuanxx-max` for DSH, which nests forked sessions under parent sessions with indentation based on `parentSessionId`. It provides rounded connection lines, green plus-collapse icons, ancestor-chain hover highlighting, and optional auto-naming/order persistence. It explains npm and link installation, `cordis.patch.yml` configuration, the division of responsibilities between Host and Client, as well as platform support and known limitations.
Read MoreEnterprise Information Query: Integrating Enterprise Data from Tianyancha, Qichacha, and Aiqicha into Conversational Interfaces
# Introduction When conducting due diligence, cleaning sales leads, or writing scripts to batch-verify a company's business registration information, the common practice is to scrape web pages manually, call third-party crawler APIs, and then write a host of regular expressions for parsing. The problem with such solutions is that they fail as soon as site structures change. Field names vary across platforms, and critical data points—such as names, registered capital, shareholders, outbound investments, judicial risks, and operational anomalies—must be pieced together separately, making error tracking difficult. The "Enterprise Information Query" Skill introduced below encapsulates such queries into a capability that can be invoked directly by an Agent. It positions itself as the "Chinese version of the Bloomberg Terminal"—essentially turning enterprise business data queries into a tool within a conversation, rather than requiring you to write crawlers yourself. # What is this? In one sentence: It is an enterprise information query Skill designed for Agents.
Read MoreDSH Background Image: Customizing the Harness Web UI Background
DSH Background Image (ASUKAq01/dsh-background-image, v1.0.0, MIT) provides local background images for DeepSeek Harness Web GUI: supports JPG/PNG/WebP/GIF formats, 0–100% opacity adjustment, and 0–40px blur. It is theme-adaptive and does not alter the layout. Installation requires cloning the repository with git clone, then enabling it via the plugin manager's local directory or by running `dsh plugin add ./dsh-background-image`. Data is stored exclusively in localStorage. Requires Harness >=0.1.7-rc.1 <0.3.0-0.
Read MoreNature Literature Downloader: Compliant Full-Text Retrieval with a Skill
Nature Literature Downloader, published by RealQ, is sourced from enterprise, categorized under data-analysis, and is version 1.0.0. It focuses solely on the full-text acquisition stage: under the premise of legal access, it covers CNKI institutional subscription access, English open-access literature retrieval, publisher API access, institutional browser fallback, and supplementary material downloads. The design logic prioritizes structured interfaces, falling back to institutional browser entry when APIs are unavailable. Page metadata indicates `requires_api_key=true`, with 1338 downloads, 0 installs, and a score of 100000, suggesting that credentials must be provided by the user and that current community adoption remains limited. The retrieved data does not include installation commands, directory conventions, or configuration fields. The text only provides the official address and metadata for verification; the usage section reiterates the trigger scenarios declared in the official introduction without fabricating parameters or cases.
Read More