Creative Frontend Visual Design: Setting an Unconventional Direction for Web Pages
Introduce the community Skill "Creative Frontend Visual Design" (/cn-frontend-design, v1.0.0, design-media category). It is designed to provide unique visual directions when creating or reshaping websites, landing pages, dashboards, UI components, posters, or applications, avoiding generic AI-generated aesthetics, with the tag . The text outlines its positioning, core features, applicable scenarios, and limitations, and honestly states that the official installation command, internal rules, and reproducible examples could not be verified during this retrieval. Both official addresses were inaccessible at the time, so no commands or usage methods are fabricated.
Read MoreEssay Coaching for High School Entrance Exams: Prompt Analysis and Writing Elevation Skills for Exam Scenarios
The Chinese high school and college entrance exam essay coach is an educational skill released by Kunlun Growth on SkillHub, located at @kunlungrowth/edu-essay-coach, version 1.2.0, source enterprise. It breaks essay correction into five stages: topic analysis, thesis formulation, structure, materials, and language, covering four scenarios: high school entrance exam essays, college entrance exam essays, essay upgrading, and essay polishing. Trigger words include essay correction, high school entrance exam essay, college entrance exam essay, essay score improvement, essay structure, and essay materials. requires_api_key=false, rating: 100,000, downloads: 1,034, installations: 0. This document outlines its capabilities, usage ideas based on the introduction, and usage notes. The official page was inaccessible during verification, and installation commands and client configuration were not provided in the materials. No information was fabricated in the text, and only the coordinates and version are provided. It is recommended to visit the official address before use to confirm.
Read Moredsh-plugin-bidkit: Native DSH Plugin for the Full Bidding Process
dsh-plugin-bidkit is a native DeepSeek Harness workflow plugin maintained by Jegozhou under the MIT license. It covers 18解析 aspects of tender documents, scoring alignment cataloging, knowledge base retrieval with fact locking, section continuation and sub-agent batching, Mermaid/SVG illustration generation, comprehensive compliance review, and pure JS layout-grade docx export, while providing a bidding workbench in the sidebar. Built on Cordis with zero external runtime dependencies, the project state is disk-persisted and recoverable. Installation is performed via `plugin_manager install_bundle` after cloning the repository, with configuration defined in `cordis.patch.yml`.
Read MoreTaobao/Pinduoduo Product Copywriting Generator: A Skill for Generating Complete E-commerce Copy in One Go
A community Skill, version 1.0.0, for generating product copy tailored to domestic e-commerce platforms such as Taobao, Pinduoduo, and Douyin. Released by user_b50eba52 and categorized under content-creation, it does not require an API key. It covers four main capabilities: product titles, detail page copy, main image copy, and promotional campaign copy. By producing a complete set of content based on product information in a single pass, it minimizes the need to repeatedly rephrase the same selling points. Existing documentation does not provide command-line installation methods, configuration fields, or official examples. The main text outlines general usage conventions regarding the SKILL.md format and highlights important considerations, including its community source, zero installations, and the necessity for manual review to ensure platform compliance.
Read Moredsh-live-spend: Real-Time Conversation Cost and Balance Monitoring in DSH
dsh-live-spend is a DSH Web client plugin maintained by jeff072333. It displays the current session's cumulative cost, DeepSeek account balance, and peak/off-peak billing periods above the input box, and shows the total consumption for each round at the end of every response. Costs for deepseek-flash and deepseek-pro are calculated based on the official price list. The balance is retrieved via `account.getBalance` using the logged-in account, requiring no API key. It requires DSH 0.2.1-rc.5 or higher. To install, run `dsh plugin --profile web add github:jeff072333/dsh-live-spend`. It only tracks the current session, and for unsupported models, it records token usage only.
Read MoreSenior Specialist in Digital and AI Transportation (Standard Edition): A Skill Covering Global Transportation Technology Selection Reference
Traffic Digitalization & AI Senior Expert (Standard Edition) is a global reference system for traffic technology solutions designed for technical professionals. Maintained by community publisher user_82981261, it is categorized as 'professional,' requires no API Key, and is currently at version 1.0.0. According to the official description, it covers 15 major transport modes, 130+ digitalization scenarios, 25 key AI applications, 140+ vendors, 12 major categories, 10 architectural patterns, and 31 knowledge domains. Key terms include Smart Traffic, Vehicle-Road Collaboration, Digital Twin, V2X, Autonomous Driving, ITS, MaaS, Smart Highways, Smart Ports, Smart Urban Rail, Signal Optimization, ETC, TCO, and ROI, with corresponding English-side frameworks such as T-DMM and RICE++. Official page statistics show a SkillHub rating of 100,000, 954 downloads, and 0 installations. It is suitable for use as a benchmark entry point during stages of technology selection, architecture design, reliability engineering, and investment decision-making. The article also clarifies its positioning as a reference system rather than a real-time data source; while it offers broad coverage, its depth in specific areas is limited, meaning key conclusions must still be verified against original standards and vendor documentation. It also candidly notes that the official source does not provide specific installation commands, and installation variations across different clients remain unverified.
Read MoreZenovaOS AI: Querying PV and Energy Storage Station O&M Data Using Natural Language
ZenovaOS AI (Coordinates @user_551019e9/zenova) is a community-published Skill, version 1.1.3, categorized under it-ops-security, sourced from the community, with the tag requires_api_key=false. Through the Zenova MCP connector, it lets the model query O&M data for PV and energy storage plants, enabling natural-language questions without opening the backend or writing scripts to glue APIs together. Verified capabilities include: generation analysis (generation and power curves, losses and theoretical generation, equivalent operating hours, availability; by default it uses meter or inverter data sources based on the plant’s energy calculation convention), alarm diagnostics (rankings, calendar, temporal distribution), device queries (inverters, meters, combiner boxes, pyranometers, loggers, cameras, cleaning robots including inspections; energy storage PCS/BMS/SOC/cabinet temperature/charging and discharging/revenue), planning and revenue (electricity prices, revenue, planned generation), knowledge Q&A (PR, string troubleshooting, communication protocols), as well as document analysis for files/images, electricity bill reconciliation, and in-depth report generation. File and image analysis must go through this connector rather than the model’s built-in vision. Topics covered include plants, generation, alarms, PR, strings, energy storage, inspections, electricity prices, revenue, planned generation, weather, availability, and communication protocols. The gathered materials do not provide specific installation commands; enabling instructions should follow the official URL and the SkillHub page. The materials do not separately describe Cursor, Codex CLI, or Claude Code environments, so no environment-specific distinctions are made. When asking questions, it is recommended to specify the plant, time range, and calculation convention; data accuracy depends on the device onboarding status at the plant.
Read MoreXinghai Academic Paper Writing Assistant: From Writing to Word Export Using a Markdown Workflow
Xinghai Academic Writing Assistant is an Agent Skill for academic writing developed and maintained by community developer user_e0723adc. Currently at version 1.7.0, it is sourced from the community and categorized under content creation. Tagged with `requires_api_key=false`, it does not require additional API keys for activation. Utilizing a Markdown-based workflow, it integrates topic planning, body writing, and Word export into a unified pipeline. It supports five document types—journal articles, review papers, dissertations, conference papers, and technical reports—and generates Word files that adhere to academic typesetting standards. Clear triggers include writing, drafting, or revising papers; planning structures or outlines; exporting to Word; literature search, citation management, and quality checks; and writing following IMRAD or other academic structures. Official sources have not provided installation commands, directories, or configuration fields, so activation methods are subject to the official address. This article only documents verified capabilities and excludes unverified details.
Read Moredsh-bili-whale-safever-by-jediliu: A safever fork of the Bilibili 'fish-farming' plugin with added entry validation
An MIT fork (0.4.6-safever.1) by jediliu005 based on dsh-bili-whale v0.4.6. It retains "fish-raising" capabilities such as scheduled deep reading, diary, and @ sharing, while adding Bilibili text input validation, stdout scanning for tools like 'debrief', a maxAgeDays publishing window with candidate filtering, and a "Safety · Entry Validation" settings page. To install, first run `npm install --omit=dev` in the plugin directory, then link the absolute path to your profile. Please read the source code and grant full permissions to the bili-whale session before use.
Read MoreHumanizing Writing Assistant for Webnovel Platforms (Fanqie, Qimao, etc.): De-AI-ifying Writing Skills for Online Literature
This is a "De-AI-ified" novel writing Agent Skill published by "Xinfeng Intelligence" on SkillHub. Versioned at 1.0.0 and categorized under content-creation, it requires no API Key. The skill covers 16 platforms, including WeRead, iReader, Douban Read, Zhihu Yanxuan, Fanqie, Qidian, Jinjiang, and QQ Reading. It codifies "platform tone adaptation + fundamental human-like writing + voice separation + long-form consistency" into writing rules. It is used for writing novels, story conception, outline design, world-building, character design, chapter content, rewriting to remove AI flavor, opening hooks, chapter-end cliffhangers, and preparing for contract signing and debuts. The human-like writing component enforces five constraints: zero banned words, varied sentence lengths, diverse dialogue styles, concrete psychological descriptions, and integrating action into environmental descriptions. The long-form consistency component tracks foreshadowing throughout the entire narrative. Installation commands and configuration fields were not provided in the retrieved official materials; the article only lists verifiable coordinates, version, and source information, explaining that the skill is enabled by placing it in the skills directory following the standard SKILL.md structure.
Read MoreOptimizing Follow-up Sales Letters & Calls-to-Action | Jianshi AI: Draft and Review Emails Based on Your Writing Style
This is a community-oriented writing Skill published on SkillHub at @user_fccf428b/email-drafter, categorized under business-ops. It leverages WorkIQ to analyze the tone, salutations, structure, and sign-off patterns of emails a user has already sent, and then drafts and reviews professional emails that align with their personal style. It is primarily aimed at customer follow-up scenarios. In the overview, the section describing generation capabilities is truncated, so specific capabilities for sales letters and CTA optimization could not be verified. The materials do not provide installation commands or configuration fields, so this article does not fabricate them. Platform-side records show 100000 rating, 909 downloads, and 0 installations.
Read Moredsh-plugin-translate: Chat translation panel and layout-preserving PDF translation
Introducing the community plugin **dsh-plugin-translate** (by jcy0chenyuan0): It provides a hover-to-translate panel at the top of the DeepSeek Harness chat area, supporting multiple modes including DeepSeek account credentials, OpenAI-compatible APIs, and Microsoft Translator. It utilizes structured slicing to protect formulas and code, and can translate text-layer PDFs block-by-block and save them as _t.pdf files. This documentation explains how to use the `install.ps1` and `install.mjs` scripts to generate a local `cordis.patch.yml` file, enable the plugin via `plugin_manager install_bundle`, and utilize the `translate_text` and `translate_pdf` tools.
Read MoreMethodology for Mathematical Popular Science Animation Video Production: A Workflow from Concept to Final Output
An introduction to the "Methodology for Creating Popular Science Animation Videos on Mathematics" on SkillHub: published by Kunlun Growth, version 1.0.0, category design-media, source enterprise, tag requires_api_key=false. The core is a six-step workflow: script design based on "one concept, one visual metaphor"; storyboarding and pacing control; selection of procedural animation tools; organizing the rendering pipeline; and compositing voice-over and subtitles. It covers four topic areas: mathematics, physics, algorithms, and AI principles, and includes a "concept-to-visualization" translation checklist along with three common pitfalls for beginners (attempting too much, showing off technical flourishes, and lacking a narrative thread). It is suitable for educators in teaching and training, creators of educational content, and technical evangelists. The article notes that the official documentation does not provide verifiable installation commands or host-environment differences; it avoids speculating on commands and provides a recommended working sequence for evaluation.
Read Moredsh-plugin-qqsend: Sending a single message to a QQ Group or Friend within DSH
dsh-plugin-qqsend is a networking tool plugin for DSH, independent from the read-only qqchat. It sends a message or a single file to a specific group or friend via a local OneBot/NapCat instance, with dry-run enabled by default. Actual sending requires multiple gates, including allowSend, confirm, and whitelist. It also provides tools for doctor checks, target resolution, scheduled tasks, and media download. To install, first run install.ps1 or node install.mjs, then install the bundle in DSH using a local path.
Read MoreShenlun Essay Skill: Integrating Material Synthesis, Framework Construction, Standardized Expression, and Grade Elevation into a Standardized Module
Shenlun is an education-category Skill published by Kunlun Growth on SkillHub, version 1.1.0, located at @kunlungrowth/kunlun-shenlun, and tagged with requires_api_key=false. It packages four stages from the Shenlun exam preparation pipeline—summarizing materials and extracting key points, structuring essay frameworks, generating standardized expressions, and grading and upgrading—into a standard SKILL.md file, enabling the Agent to process Shenlun tasks via a fixed workflow and reducing improvisation. Since the official page does not provide specific installation commands or parameter fields, this article only introduces the verified positioning, capabilities, and activation methods. It is recommended to confirm installation details on the official page before attempting to use it.
Read Moredsh-plugin-qqchat: Read-Only Extraction of QQ NT Local Chat History
Introduces the DSH plugin `dsh-plugin-qqchat` (v1.0.0, admin-security) by `jcy0chenyuan0`. It performs read-only decryption of local QQ NT databases on copied snapshots and provides tools such as `qq_doctor`, `qq_find_key`, `qq_decrypt`, and `qq_media`. It does not inject into the QQ process nor send messages. The text also explains how to enable the plugin via `install.ps1` on Windows and `plugin_manager install_bundle file:` paths, as well as planned features currently in progress, such as conversation export.
Read MoreReplying to Colleagues: Compressing Workplace Message Replies from "Overthinking" to a Single Confirmation
"Reply to Colleague" is a community Skill on SkillHub designed for workplace message reply scenarios. It supports six use cases: leave requests and approvals, task delivery, meeting conflicts, progress updates, collaboration requests, and expressing gratitude. It works with instant messaging tools such as WeChat, WeCom, and Microsoft Teams. Activated by Chinese and English trigger phrases (e.g., "帮我想想怎么回", "回领导", "draft reply"), it generates ready-to-use message drafts but does not send messages itself. No API key is required. Current version: 1.0.2. Category: office-efficiency. The article outlines its positioning, activation methods, applicable scenarios, usage notes, and notes that installation steps should follow the official page.
Read MoreShuidi Credit Enterprise Due Diligence - Light Due Diligence Prior to Partnership: The Skill of Vetting Partners Before Signing Contracts
This article introduces the community Skill “水滴信用企业尽调-合作前的轻尽调” (located at @user_a15e4142/enterprise-due-diligence-shuidixinyong; version 1.0.7; category: business-ops; tag: requires_api_key=true). It is a corporate due-diligence assistant from a bidding and tendering perspective: given a company entity, it generates a lightweight due-diligence report based on cross-web bidding and tendering data—operational activity is evidenced by winning bid records; customers and suppliers are shown as real contractual relationships; competitors are derived from bidding overlaps; contract-performance capability and reliance on key clients are included in the assessment; publicly disclosed litigation and administrative penalties are listed item by item with source links. It supports in-depth reports for a single company and side-by-side comparisons between two companies. Typical use cases include partner vetting before cooperation, signing, or payment-term decisions, customer background checks, verification of actual operations, and risk screening—even when the user does not explicitly mention “due diligence.” The article also outlines the division of labor among this Skill and the three Skills zlbx-bid-decision, zlbx-opportunity-radar, and zlbx-bidding, notes that the data relies on bidding/tendering records and that coverage is limited to publicly available risks, and provides the official URL.
Read MoreOutlook Email Sorting Assistant: Automated Email Classification Based on the Graph API
The Outlook Mail Triage Assistant is a community-sourced Agent Skill in the office-efficiency category, located at @user_3c6cb52e/outlook-mail-triage-assistant, published by user_3c6cb52e. Version 1.0.0 requires an API Key. The official description states that it uses Microsoft Graph API and @azure/msal-node for automated email classification, performs zero-shot classification on messages, applies Outlook rules, and manages Focused Inbox priorities. The record shows a rating of 100000, 985 downloads, and 0 installs. The article verified the official page and noted that the official source does not provide complete installation commands or the SKILL.md content; before enabling it, Azure AD app registration, Graph permissions, and an API Key must be prepared. The typical workflow involves authentication, reading, zero-shot classification, applying rules, and adjusting priorities. It is recommended to first validate in a test mailbox, and to be aware of the community source, permission scope, and the risks of the early 1.0.0 version.
Read MoreBooGo AI HTML to PDF: Convert Online Slides and Axure Prototypes to PDF
This article introduces the community skill "Boo哥AI-html转pdf" (@user_3c9003af/boo-ai-html-pdfv1, version 1.0.0, category office-efficiency). It targets webpage-to-PDF scenarios: online prototypes (axureshow.com etc.), HTML slides, screenshot-based composition of single-page presentation pages, and exporting web pages with dot or thumbnail navigation. It is triggered by keywords and requires no API Key. The official page was inaccessible during this check, so the installation command and configuration fields are not included; only verifiable coordinates and version information are provided. It also reminds users that conversion quality is affected by page styling and spot checks should be performed before export. Ratings and download counts are based on the current page.
Read Moredsh-plugin-python: Adding Model-Callable Python Tools to DSH
dsh-plugin-python is a DSH profile bundle maintained by jcy0chenyuan0. It registers Python tooling with Cordis and executes model-submitted code using Harness-bundled Python 3.12.14 with the numpy, pandas, and Office-related libraries. According to the README, you use install.ps1 or node install.mjs to generate the patch, then run plugin_manager install_bundle to install it; the article documents the code, files, workdir, and timeoutMs parameters along with the sandbox policy. Suitable for developers who need to run Python within sessions and want alignment with the DSH runtime; please review the source code and the MIT license before using.
Read Moredsh-plugin-open-in-app-fix: Fix DSH "Open In" failure to open Electron editor
During a DSH session, using Open In to launch Electron editors such as VS Code, Cursor, or Windsurf often causes the child process to exit immediately and report "Open failed, please retry" because the Desktop host inherits ELECTRON_RUN_AS_NODE=1. The plugin jcy0chenyuan0/dsh-plugin-open-in-app-fix runs inside the host process and filters out this variable only for spawns that are detached with stdio ignored, so it does not affect shell or Python child processes. To install it, first run install.ps1 or node install.mjs to generate cordis.patch.yml, then execute plugin_manager install_bundle file:<absolute-path>. An optional run-selftest can be performed; the fix is effective if probe.outcome is launched.
Read MoreParenting Guidance: Age-Specific Scripts & Response Strategies
A parenting consultation Skill organized by age (monthly age/age), maintained by community publisher user_8ee65e73 and identified by the coordinate @user_8ee65e73/parenting-coach. The current visible version is 1.2.7, categorized under life-service, and does not require a mandatory API Key. It focuses on four scenarios: feeding and sleep, emotional guidance, early education games, and preschool transition, providing actionable activities and key parenting points while emphasizing a non-preachy approach that does not create anxiety. To enable it, the coordinate, name, and category must be declared in the host. If used in tools such as Cursor, Codex CLI, or Claude Code, the parenting-coach coordinate should be configured according to each tool's Skill loading method. The official description includes mandatory routing rules: if a user mentions parenting, childcare, feeding and sleep, emotional guidance, early education games, or preschool anxiety, or pastes original text/links/screenshots, this Skill must be invoked immediately rather than asking clarifying questions first or substituting with general capabilities. Before integration, it is recommended to self-test trigger stability and the granularity of age-based outputs. Note that there are discrepancies in the rating and download data displayed on the page; do not treat these as effectiveness metrics.
Read MoreNuwa Creates Humans: Inputting a Name or Topic to Automatically Distill Thinking Frameworks and Generate Runnable Persona Skills
Nuwa Creator (@user_814dbe54/x-nuwa-creator, version 1.0.1) is an ai-agent type Skill published on the SkillHub community, designed to automate the most time-consuming step in the Skill creation process: extracting mental frameworks from person names or topics to produce directly runnable character Skills. It is triggered when users need to create Skills, distill a person, acquire a certain thinking mode, or require a thinking advisor. It is marked with requires_api_key=false, meaning no API Key is required. This article, organized based on the official introduction, outlines its positioning, core capabilities, installation entry point and activation methods, typical trigger usage, and explains applicable scenarios and caveats: it is a generator rather than a knowledge base, the more specific the input, the more valuable the output, and generated frameworks require self-review; it also notes that the official documentation does not provide CLI commands, directory structure, or reproducible examples, the installation count is 0, and its quality needs to be verified through empirical testing.
Read MoreEfficient Learning Framework Consultant: Transforming Passive Reading into Active Inquiry
Presentation of the "Efficient Study Method Framework Consultant" Skill on SkillHub This article introduces the "Efficient Study Method Framework Consultant" skill available on SkillHub. Located at @kunlungrowth/study-method-mcmurry, it is published by Kunlun Growth as Version 1.0.0. Classified under education, the skill is standalone and requires no API key. Grounded in the "eight key elements of learning" from McMurry's *How to Study and Teaching How to Study* (1909, a public domain text by Project Gutenberg), this skill targets common study issues such as forgetting material shortly after studying, struggling to grasp key points, and poor note-taking. It also addresses passive reading and inefficient planning, as well as focus and concentration issues. It transforms reading, attending lectures, exam preparation, and on-the-job self-study from passive reception into a purposeful, structured, and reusable active research process. The article outlines the skill's positioning, core capability directions, official web address, and activation mechanism. It further notes that the extracted materials did not provide specific installation commands or reproducible examples; therefore, no commands are provided in this text, and users are advised to check the official page for accurate instructions. It is also highlighted that, since this methodology originates from a 1909 publication, it should be utilized as a structural framework rather than the sole conclusion. Additionally, the current installation count of 0, 965 downloads, and a rating of 100000 are data points as claimed by the publisher and do not necessarily represent large-scale verification.
Read More