IELTS/TOEFL Prep: Delegating the Four-Section Study Plan, Scoring Rubrics, and Mock Exam Review to an Agent
Published by Kunlun Growth (located at @kunlungrowth/edu-ielts-toefl), this Skill is version 1.0.0, categorized under Education, sourced from Enterprise, and requires no API Key. It deconstructs IELTS and TOEFL preparation into four steps: planning, understanding scoring, gathering materials, and reviewing. It covers the interplay between the four sections—Listening, Speaking, Reading, and Writing—rather than merely providing a general timetable. This article outlines four verified capabilities, provides four sets of invocation examples at the capability level, and candidly notes that the official page was inaccessible and the installation command was unverified. Metadata shows 917 downloads and 0 installations; it is recommended to verify this on the SkillHub page before adoption.
Read Moredsh-session-temp-reaper: One-click cleanup of session temporary files
This document introduces the `dsh-session-temp-reaper` (admin-security) plugin, maintained by liyolin for DSH. It provides a cleanup button next to the input box, utilizing `mark_temp_files` to register temporary files via the model. The panel categorizes files into "marked" (selected for deletion by default) and "unmarked" (retained by default). Deletion requires double confirmation, and the Host side protects deliverables and profile dependencies. The document includes GitHub installation commands, bundles configuration, typical usage scenarios, and an explanation of security boundaries.
Read MoreStandardized Documentation Drives Code Development: Finalize the Documentation First, Then Write Code
This community Skill (@user_2bb1379b/structured-dev-flow, v2.0.1) enforces a rigid nine-step development workflow: open-source research → PRD → interaction documentation → system design documentation → coding → functional testing → UI interaction testing → human-machine interaction validation → delivery and packaging. Documents for each stage must be confirmed by the user before proceeding to the next. Comments are mandatory for the interface and Service layers; synchronous document updates, minor version recording, and regression testing are required when fixing bugs. It is suitable for new projects built from scratch and medium-to-large scale feature development, but is not applicable to one-line scripts or purely Q&A interactions.
Read More0177: Dark Purple and Gold Themed Website Design System
0177 is a dark purple and gold themed website design system Skill reconstructed from the Photofolio photography portfolio template. It is maintained by SkillHub community publisher user_6ca8e607, classified under design-media, requires no API key, and is at version 1.0.0. This Skill includes a complete static template site and design specifications, suitable for tribute sites, personal portfolios, photography galleries, sports hall of fame websites, and web designs that explicitly require a purple and gold color scheme, a dark premium aesthetic, or Mamba/Lakers styling. The article introduces its positioning, verified core features, installation access, and typical usage, while also highlighting considerations such as static site extensions and license verification.
Read Moredsh-voice-studio: Managing GPT-SoVITS and RVC Covers in the DSH Sidebar
dsh-voice-studio is a Windows-only DeepSeek Harness client plugin maintained by liye0803. It manages GPT-SoVITS zero-shot audition, training, and voice/timbre switching in the sidebar, with optional RVC cover functionality (source separation, voice conversion, and mixing). The plugin does not include models; users must provide their own Harness, api_v2.py, and an optional RVC environment. It offers an environment self-check, engine keep-alive, and a GET /voice-studio/say HTTP endpoint. Installation: dsh plugin --profile desktop add dsh-voice-studio. MIT licensed, v1.0.2.
Read MoreShort Video Scriptwriting (TikTok/WeChat Channels): Breaking Down a Video into a Ready-to-Shoot Script
This article introduces the community Skill "Short Video Script Creation (Douyin/Video Accounts)"@user_4f99260d/short-video-script-v3 (Version 1.0.0). Sourced from the community and categorized under content-creation, it requires no API Key. This Skill expands requests such as "write a short video script" into complete deliverables covering voiceover scripts, storyboards, scene designs, dialogue, BGM suggestions, and shooting guidance. It supports four script genres: knowledge-based, product promotion, narrative, and vlog, targeting platforms including Douyin, WeChat Channels, Kuaishou, and Xiaohongshu. The article provides four typical query examples and outlines the suitable audience and limitations: the format of output fields needs to be validated through actual testing; platform-specific duration and "hook" rules require manual judgment; and since the official page shows zero installations, indicating a newly launched Skill, it is recommended to test with a single script on a small scale before batch usage. As public information does not provide installation commands or the SKILL.md content, the article makes no speculations on these aspects.
Read Moredsh-active-chat: Pinning Active Session Workspaces to the Top of the Sidebar
@lixiangzhong/dsh-active-chat is a DSH Web client plugin that, based on the same "in progress" signal as the sidebar activity dot (running or waiting for your reply), automatically moves workspaces that still have active conversations to the top of the Host sidebar; when all are idle, it stops reordering and preserves the manual order. Requires DSH >= 0.2.1-rc.2. CLI install: `dsh plugin --profile web add @lixiangzhong/dsh-active-chat`. No configuration required; enabled immediately upon installation. MIT licensed.
Read Moredsh-wechat-relations: Bringing Contact Relationships into the DSH Sidebar
`dsh-wechat-relations` is a relationship analysis plugin for DeepSeek Harness maintained by liuzhuomin. Its sidebar offers network, timeline, insight, classification, and individual profile views, where health and care scores are expressed as percentiles within the current batch of contacts. The repository contains only demo data generated by `mock-data`. To enable it, clone the repo, run `mock-data.mjs` and `build.mjs`, then mount the local directory using `dsh plugin add`. For real chat data, you must provide your own JSONL or use an external analysis script. Decision recommendations default to the DSH model, but can be configured via POST to JEV. Licensed under MIT; calculations are performed locally, though the decision-making process sends the most recent 1200 chat messages to the configured model.
Read Moredsh-art-workshop: Video and Image-text Production Pipeline in DSH
dsh-art-workshop is a DeepSeek Harness plugin maintained by liuzhuomin that decomposes the workflow into replaceable chains covering topic selection, scripting, voiceover, visuals, composition, review, and publishing. It provides five views—Works, Chains, Plugins, Assets, and Knowledge Base—and supports chain-config persistence, plugin self-testing, and token/time consumption statistics. It can be installed by cloning the repository and running `dsh plugin add .`, with optional environment variables to specify the Remotion project and output directories. It is suitable for developers needing to orchestrate multiple DSH content plugins; please review the MIT source code before use and pay attention to security regarding current process privileges.
Read MoreLife & Entertainment Guide for the Changlong Island Campus of Hubei Institute of Fine Arts: Turning "Where to Eat Near HIFA and Which Traps to Avoid" into a Callable Decision Skill
This is a Knowledge Base and Decision-Making Skill for shop recommendations, avoidance tips, and a Lifestyle Guide around the Longdaodao Campus of Hubei Institute of Fine Arts (HIFA). Published by user_b332ace1, sourced from the community, categorized under life-service, version 1.0.3, with the tag requires_api_key=false. It is designed for developers and Agent applications that need to answer questions about dining options and daily living around HIFA in conversational contexts. The core capabilities include three main features: 1. Triggering by geographic location and lifestyle requirements, covering areas such as HIFA, Longdaodao, Meiyuan Village, Limiaoxincun, the snack street, South Gate 2, Hubei University of Economics, and Yuxian Zhai, as well as dining, store exploration, food delivery, avoidance tips, accommodation, hair salons and beauty services, study spaces, and sports and fitness; 2. A two-tiered rating system that compares similar shops using P/E/S three-dimensional scoring, while preserving irreplaceable options with a U (Uniqueness) rating; 3. Demand profile matching, which provides tailored recommendations for different users, along with precise shop locations and photos of storefronts. Regarding installation, the Skill is based on the general SKILL.md format and does not require an additional API Key. However, official documentation does not provide specific installation commands or platform-specific differences for Cursor, Codex CLI, or Claude Code, and can only be obtained through official links and the SkillHub page. The article also provides four reproducible question examples, applicable scenarios, and notes on content scope, undisclosed scoring details, and potential page scraping failures.
Read More"Financial Report Writer": A Skill for Auto-Generating Structured HTML Financial Analysis Reports
This article introduces the SkillHub community Skill "Financial Report Writer" (@user_053d27d0/wzx3333, version 1.0.3, category: data-analysis, source: community). It automatically generates structured HTML financial analysis reports and outputs them in document format when users request financial reports, stock analyses, investment reports, or market overviews, without requiring an API key. The article outlines its core functions, official address and coordinates, and typical triggering scenarios. It also points out that the source has not disclosed installation commands, data sources, or template details. Since the actual installation volume is relatively low, it is recommended to perform a small-scale validation of the output quality before use.
Read Moredsh-token-rmb: Inline display of per-turn RMB token estimation in DSH operation lines
dsh-token-rmb is a Web plugin for DeepSeek Harness that displays the RMB estimate of token costs for the current round, positioned adjacent to usage and time information in each assistant operation row. It automatically differentiates between peak and idle pricing rates, does not add new rows, and can be hidden together with the "Performance & Usage = Compact" view. Usage data is sourced from the same origin as the official panel, and pricing can be entirely overridden via the profile's cordis.patch.yml. This plugin is suitable for developers requiring lightweight inline cost indicators; it does not handle statutory holidays and lacks session accumulation, representing different feature trade-offs compared to dsh-turn-cost. A profile restart is required after installation, and it is compatible with dsh versions from 0.2.1-rc.3 up to (but not including) 0.2.2.
Read MoreAI Short Drama: Production Skills from Script Review to Video Generation
AI Short Drama (@user_db7424f5/udb7-ai-short-drama, version 1.3.1, community source, category: design-media) is a Skill designed for the entire AI short drama production workflow. It consolidates five steps—script review, character design, storyboard breakdown, first/last frame keyframes, and video generation—into a single SKILL.md instruction package. It covers genres such as villain prequels, aiming to eliminate the need for renegotiating format conventions each time for short drama projects within the same IP. The materials only provide distribution address, coordinates, and version information, without including installation commands, directory paths, or configuration fields; this article does not infer installation methods. The tag requires_api_key=false indicates that no additional API Key is required per this tag. The page shows 900 downloads and 0 installs, with a rating field value of 100000; such maximum scores originate from the platform's own scoring mechanism and should not be regarded as evidence of usage effectiveness. As a community-maintained Skill, it is advisable to validate storyboard granularity and first/last frame selection methods on small sample scripts before integration; the quality of video generation still depends on external capabilities.
Read MoreLevel Design: Deconstructing Game Mechanics into Implementable Skills
This article introduces the community Skill "Level Design," published by user_3c6cb52e, with a community source, professional classification, and version 2.0.2. It conducts an in-depth exploration centered on level design, covering both game mechanism analysis and competitive strategies. The trigger word is "level-design," and the tags indicate no additional API Key is required. The article starts with the real pain points of level designers, explains the Skill's positioning and three verified capabilities: mechanism analysis, level design, and competitive strategies. The installation and activation section honestly notes that the official materials do not provide specific installation commands or configuration fields; it should be handled according to the client's method for loading community sourced Skills, without copying non-existent commands. Typical usage provides three minimal call examples corresponding to mechanism, level, and strategy respectively, with notes on writing considerations. Finally, it lists applicable scenarios and limitations, including the lack of community feedback due to 0 installations, its role in providing analysis suggestions only without generating resources, and the fact that data such as ratings and download counts are subject to real-time values on the official page. The article ends with the official address, SkillHub page, and coordinates.
Read Moredsh-spanish-tutor: DSH Spanish Practice and Local Scoring
dsh-spanish-tutor is a Spanish practice plugin for DeepSeek Harness maintained by liuyize-CHE. Exercises on verb conjugation, noun gender and plural forms, vocabulary recognition, fill-in-the-blank, and specialized grammar are generated and scored by a local rule engine, while situational sentence creation is graded by the model. It supports learning before practicing, SRS (spaced repetition system), and persistent progress. The installation command is `dsh plugin --profile web add github:liuyize-CHE/dsh-spanish-tutor`; a DSH restart is required after installation. The model is mainly used for explanations and free expression, while scoring is recomputed on the host side to prevent cheating.
Read MoreBid Generation Expert: Skill for Structuring Bids According to Evaluation Rules
"Bid Document Generation Expert" is a Skill published by SkillHub community author user_18cb0f1a (Version 1.0.1). Classified under business-ops and sourced from the community, it currently requires no API Key configuration. It targets procurement across three major categories—engineering, goods, and services—covering the bidding logic and evaluation rules for projects involving government procurement, state-owned enterprises (SOEs), construction, IT software, and consulting services. Its function is to deconstruct scoring criteria, construct chapter structures, and refine response phrasing, rather than generating generic copy. Official scraped data did not provide directly reusable installation commands or invocation examples, and both relevant pages failed to load. Therefore, this article accurately states the premise regarding installation coordinates and the lack of key requirements, prompting readers to visit the official address to verify installation instructions. As this Skill originates from a community maintained by a personal author, it is mandatory to cross-check it against the tender documents item by item before formal bidding; the risk of bid rejection lies solely with the bidder.
Read MoreEducational Shrimp 15: Age-appropriate Holistic Companion Skill in OpenClaw Course Session 15
Education Shrimp 15 is an education-category Skill published by community author user_d93c4aed on SkillHub, located at @user_d93c4aed/sc-edu-growth-companion, version 1.0.0, with tag requires_api_key=false. It is used for OpenClaw Course Lesson 15: age-graded moral, intellectual, physical, and aesthetic companionship, parent summaries, and suggestions for learning/exercise/aesthetic tasks. Trigger words: education, child, parent, moral/intellectual/physical/aesthetic, age-graded, growth, companionship. Page snapshot rating 100000, downloads 927, installs 0. This article describes its positioning, core functions, official activation address, trigger-word usage, applicable scenarios, and precautions.
Read Moreskill-learning-path: Turning "Want to Learn a Skill" into an Actionable Learning Path Report
skill-learning-path is a community-sourced Skill in the education category, version 1.1.0, published by user_38e401bd, and does not require an API Key. It generates detailed learning roadmap reports based on the skills users want to learn, covering learning direction, phase division, timeline planning, and specific resource recommendations. The officially listed trigger phrases include "Help me create a learning plan," "I want to learn XXX, how do I get started," "Recommend a learning path for XXX," "How to learn a particular skill," as well as scenarios where understanding learning cycles and resources is required. The output is a structured Markdown document that can be further edited. The article provides the installation address, reproducible requirement examples, and applicable boundaries; since public page materials do not provide specific installation commands, platform configuration differences, or resource filtering rules, no unverified extensions are made.
Read MoreLao Mo Bid Review Master: 62 Rigorous Checks and Self-Inspections for Preliminary Bid Drafts
Laomo Biaoshu Review Master (lm-biaoshu-shenhe) is a community Skill on SkillHub (Version 1.0.0, Category: office-efficiency) that requires no API Key. It is positioned as a comprehensive self-check and review tool to be used after the initial draft of technical or commercial bids is completed. It features 62 automated hard checks covering red lines for bid rejection, anonymous bid identity leaks, format and metadata issues, data consistency, validity of normative citations, AI writing tone, and consistency between graphics/texts and across different bid sections. It includes a built-in review focus library covering 16 engineering specialties, loaded dynamically as needed. The tool supports multi-section anti-collusion checks (team comparison, scanning for words from other bid sections), OCR-based empirical verification of certificates, and cross-verification of performance record contradictions. Defects are output based on seven elements: location, original text, type, severity, basis, rectification suggestions, and relevance, categorized into levels P0, P1, and P2. Compliance and bid rejection judgments are based strictly on the specific project's tender documents, without using hard-coded default values. The base version contains general rules and scripts but excludes proprietary enterprise experience; extension methods are detailed in the "references/扩展指引.md" file within the package. The article describes the inspection scope, empirical verification capabilities, seven-element output format, trigger words, and typical usage, noting that items not verified in the documentation, such as installation commands, should refer to the official page.
Read Moredsh-workspace-detached: Pop out workspace navigation to a standalone Windows window
dsh-workspace-detached is a DSH client plugin (0.2.3 preview version) maintained by liushimu. On Windows x64, it uses public interfaces to pop up workspace and session navigation into independently movable windows, retaining overview, focus, grouping, and logical drag-and-drop sorting. Session actions are still handled by the main window. To install, download the tgz file from Releases. It cannot be enabled simultaneously with Better Workspace. When the host stops, local packages can be added with the command dsh plugin --profile desktop add.
Read MoreInstructor Da Feng's One-Stop Short Video Auto-Creation: An Automated Skill from Topic to Final Cut
Introduce the MoneyPrinterTurbo-based one-stop short video auto-generation skill, located at @indiv-dfteacher/fmg-1ai-video, version 1.3.4, in the community category under the design-media domain. It automatically handles installation, configuration reuse, generation, and final MP4 delivery from a topic or script, supporting Chinese vertical voiceover short videos and covering types such as short-form, educational, and marketing. Core features, notes on required API keys, activation methods, and typical usage are listed, making it suitable for creators and developers who want agents to automatically produce vertical short videos.
Read Moredsh-baidu: Search Engine Simulator in DSH Sidebar
dsh-baidu is a demo client-side plugin for DeepSeek Harness, providing a "Search Engine Simulator" entry in the web sidebar. It invokes `ctx.web.search` via `GET /dsh-baidu/search` to display the top 10 real results. Based on the official README, this article explains the installation commands, usage, routing, and credential borrowing mechanism, and also notes the search service configuration and plugin process permissions.
Read MoreInvestment Decision Officer: Converging the Investment Decision Pipeline into a Verifiable Decision Structure
Investment-Decision-Officer is produced by Zhihe Island Group. Version 9.6.0. It is published on SkillHub under the Professional category, at @user_ec920cef/investment-decision-officer. The skill adopts a two-layer architecture: a Rules-of-Engagement (ROE) layer and eight engines, with decision prototypes adapted to institutional styles layered on top and a three-tier calibration closed loop. The ROE layer sets red lines first—answering what it trusts and what must not be touched. The eight engines cover how to think, how to judge, how to veto, how to produce numbers, how to write committee materials, how to train instincts, how to read the three statements, and how to judge early-stage projects when there are no operating data. Coverage includes cognitive biases and due-diligence falsification, confidence iteration, cycle awareness, signal-vs-noise separation, pricing power, free cash flow, business-model flywheels, end-game of the track, founder assessment, one-vote veto and ten risk-control principles, valuation anchors, financial fraud detection, technical due diligence, SPA terms and valuation adjustment mechanisms, and cross-border structures. Every output includes five fixed elements: Conclusion, Evidence, Confidence level, Flip conditions, and Boundary statement. When materials are incomplete, it returns “Pending” (lacks information) and provides how to supplement; it does not issue a negative conclusion simply because one piece of information is missing. Deliverables include an eight-chapter investment decision memo, a one-page investment opinion for early-stage projects, and a four-question scorecard for follow-on investment, with a handover note for cross-skill referrals. Official metadata: rating 100,000, downloads 940, installs 0; tag requires_api_key=false. The article explains its positioning, core functions, installation entry, and typical usage, and lists boundaries: it does not issue formal investment advisory reports, does not promise returns, and does not produce stamped appraisals or valuation reports; legal effectiveness, applicability of accounting standards, and specific tax-saving plans are referred to licensed professional institutions.
Read Moredsh-beads-panel: Managing Beads Work Items in the Harness Sidebar
dsh-beads-panel is a DSH client plugin maintained by LiuPack that reads and writes to the workspace `bd` database via the Beads panel on the left side of the Web GUI. It supports Ready/In progress/Open/All filtering, details, creating new beads, as well as Claim/Close actions. The Host (index.js) calls `bd` via an authenticated API, while the Client (client.js) registers the sidebar entry. Beads CLI must be pre-installed; the npm installation command is `dsh plugin --profile <profile> add dsh-beads-panel`. The workspace requires `bd init`; write operations are queued; the UI supports both Chinese and English. MIT License, Node >= 22.
Read MoreQwen Image Generation Skill: Invoking Qwen-Image to Generate Images with Chinese Text
Introduces the community-sourced Skill “Qwen Image图片生成” (@user_0064faa7/moark-qwen-image, version 1.0.0, category design-media). It calls the Qwen-Image model on ai.gitee.com to generate images, supports Chinese text-to-image, text layout, negative prompts, size, seed, and LoRA, and is suitable for generating images with Chinese text such as couplets, posters, and illustrations. The tag requires_api_key=true means an API Key must be configured before use. The article explains the scope of capabilities, prerequisites for enabling, typical usage elements, and applicable limitations, and provides the official URL; because the official source does not provide installation commands or example parameters, the relevant sections state that these have not been verified, without speculation.
Read More