AI Assistant for Business Planning and Profit Safeguarding of New Real Estate Projects (Peng Ge Liao Gong AI): Advancing Project Operations to the Pre-construction Phase
This is a community Skill (v1.0.0, business-ops, no API key required) produced by "Peng Ge Liao Gongcheng AI." Designed for new construction projects in real estate, housing construction, construction management, and EPC general contracting, it covers contract risk identification, target cost locking, scanning of ten key commercial strategies (including integration of temporary and permanent facilities, earthwork and site preparation, subcontractor procurement, variation and visa management, and settlement and payment collection), as well as process validation and profit realization. It establishes six operational baselines and creates a management closed loop through a five-step process (input data, goal definition, risk scanning, special planning, and process tracking), outputting commercial planning reports, 90-day pre-commencement action plans, special planning cards, risk maps, and monthly operational reports. The article provides coordinates, official addresses, fixed input fields, and three official examples, while noting that this tool does not replace contract review, cost consulting, or legal advice.
Read Moredisplay-api-balance: Display DeepSeek API balance below the DSH input area
display-api-balance is a DSH client plugin maintained by river291 that displays the DeepSeek API balance in the composer statistics row, refreshing every 60 seconds and updating immediately when the window is focused. Clicking allows setting a local low balance threshold (default ¥20), stored in localStorage. The API key is called by the host via api.deepseek.com/user/balance and is not exposed to the browser. The installation command is `dsh plugin --profile <profile> add github:river291/display-api-balance`, requiring DEEPSEEK_API_KEY configuration, verified on DSH 0.2.1-rc.5 desktop version. MIT licensed.
Read MoreCostume Design: Skills for Film and Theatre Projects
"Costume Design" is a community-class Skill on SkillHub, located at @user_3c6cb52e/self-ent-film-costume-design. Maintained by user_3c6cb52e, it is categorized as professional (Version 2.0.2), requires no API Key, and is triggered by "costume-design." Centered on costume design for film and drama characters, the content is produced from the perspective of a film school-trained critic and screenwriter. It emphasizes the understanding of camera language, narrative structure, and blocking. The goal is to provide actionable costume solutions within text conversations, rather than relying on vague adjectives like "vintage" or "glamorous." The article is structured around positioning, core functions, installation and activation, typical usage, applicable scenarios, and precautions. It should be noted that the official page fetch failed during this process; only the introduction and metadata were retained from the retrieved materials. Consequently, this article does not include specific installation commands for Cursor, Codex CLI, or Claude Code, nor does it fabricate the phase divisions, output templates, or official examples contained within the SKILL.md. The examples section merely transforms trigger words and introduction dimensions into reproducible inputs, explicitly labeled as non-official examples. The target audience includes film and drama creators who require alignment between character styling and camera blocking, as well as developers seeking to integrate costume design reasoning chains into conversational tools. The limitation is that as a prompt-based Skill, it cannot replace professional costume design judgment and does not cover pattern making, fabric sourcing, or image generation.
Read MoreTypography and Font Design: Font Pairing, Hierarchy, Readability, and Loading Optimization
"Typography and Font Design" (*@kunlungrowth/typography-guide*, v1.0.0, published by Kunlun Growth, categorized under design-media, sourced from enterprise, no API key required) is a Skill designed for front-end development and design delivery, covering four areas: font pairing, hierarchy, readability, and web font loading optimization. The article explains its positioning and these four capabilities. It notes that the official description does not break these down into an instruction list, nor does it provide verifiable installation commands or example code. Therefore, users must consult the official page to verify installation and usage, rather than copying commands from other Skills. The article also outlines three common pain points in existing page layouts, applicable scenarios, and precautions. It is intended as a decision-making reference during the page structure writing phase, rather than a one-click configuration template to be applied blindly.
Read MoreDSH-Computer-Use: Enabling DSH Agents to Perceive and Control the Real Desktop on Windows
Introducing the community plugin `@restregular/dsh-computer-use`: powered by a built-in Rust helper, it provides DeepSeek Harness with tools such as `computer_screenshot`, `computer_action`, and `computer_windows` to operate on a real Windows desktop under manual approval and a screen takeover mask. Includes official installation commands, recommended invocation loops, coordinate space explanations, and version requirements for Windows x64 and DSH.
Read MoreOvernight Global Market Review: A Five-Dimensional Workflow
This article introduces the community skill "Overnight Global Market Review" (located at @user_fa9f97cf/overnight-global-market-review, version 1.0.0, category: data-analysis, source: community, publisher: user_fa9f97cf). This skill addresses the pre-market pain point of manually gathering overnight global market data with inconsistent data formats. Invoked via trigger words such as "overnight review," "US stock close," or "global market performance," it collects market data and news across five key dimensions: the three major US stock indices, A50/HSI futures, commodities, USD exchange rates, and critical domestic/international news. It then generates a self-contained, single-page HTML review report. No API key is required, and it can be used independently or as a submodule within the `a-share-daily-review` skill. The article outlines its positioning, core features, installation methods (since official materials do not provide specific commands, it is advisable to consult the official page), typical trigger word usage, applicable scenarios, and key considerations.
Read MoreDSH-Vantage: Overview of Dialogue-Side Read-Only Tasks
dsh-vantage is a DSH client plugin maintained by regrevia, providing a read-only task overview alongside the session: one to three work themes for the current turn, expandable evidence, verification status, and bounded history. It does not participate in scheduling or completion decisions. After installing and reloading the plugin via install_bundle, open vantage in the view switcher. The per-session toggle is off by default; once enabled, it records and optionally generates model summaries. Requires DSH >=0.2.1-rc.3; data is stored locally and the software is licensed under MIT.
Read MoreSupplier Risk Early Warning: Enterprise Risk Profiling Skill for Supplier Onboarding Scenarios
This article introduces the community Skill "Supplier Risk Early Warning" (供应商风险预警). Based on verified metadata from SkillHub (Coordinates: `@user_55a6792f/corporate-risk-local`, Version 1.0.0, Source: community, Category: professional, Rating: 100000, Downloads: 1187, Tag: requires_api_key=false), and noting that the official page was inaccessible and the public description consists of only a single line, this article focuses exclusively on verifiable facts. Details such as installation commands, scoring dimensions, and data sources are accurately marked as "to be verified." It is recommended to read the original SKILL.md file before use.
Read MorePresentation Narrative Coach: Organizing Scattered Materials into Slide Outlines, Speech Drafts, and Cut Lists
"Reporting Narrative Coach" is a community Skill published by user_bd3231bd on SkillHub (version 0.1.0, category: content-creation, tags: requires_api_key=false). It handles the pre-PPT generation workflow: transforming scattered presentation materials into a paginated outline (one argument per page), page-by-page spoken scripts, and a reduction checklist, without producing the actual PPT file. It is suitable for scenarios where the presentation logic is unclear, a spoken script is required, or time and page constraints necessitate identifying content for deletion. The output quality depends on the input materials, and the scripts require a final review to align with personal tone.
Read MoreQuantitative Trading API Skill Library: Packaging SKILL.md Skills for AI Agents
Quantitative Trading API Skills Library (@user_3c6cb52e/alpaca-skills-ruz4, version 1.0.0, community-sourced, categorized as professional) is a SKILL.md-format skill designed for AI Agents, aiming to help Agents efficiently complete tasks related to quantitative trading APIs. The page is tagged with requires_api_key=true, indicating that an API Key must be prepared before execution. The materials do not provide installation commands, configuration fields, or reproducible examples, which are honestly noted here without any speculation. The article explains the skill's positioning, verifiable capability boundaries, pre-loading confirmation requirements, and calling workflows, as well as precautions such as community sourcing, low adoption (0 installations, 1148 downloads), version 1.0.0, and the need for manual review of trading operations, and provides the official address for further verification.
Read MoreBaidu Search Ads Expert: Unify Ad Copy, Creatives, and Landing Page Optimization into One Skill
This article introduces the community Skill "Baidu Search Ads Expert" (@user_4c24e814/baidu-search-ads, v1.2.0, business-ops). Designed for Baidu Search Promotion scenarios, this Skill generates creative copy (titles and descriptions) compliant with Baidu's official creative quality standards based on keywords and landing pages. It also provides material improvement suggestions and landing page optimization plans. All three outputs follow a unified standard and require no API Key. The article explains the Skill's core functions, trigger conditions, installation and activation methods, and typical usage. It also notes that this is a community-sourced tool rather than an official Baidu product, and that features such as targeting strategies and conversion bidding have not been verified in the provided documentation.
Read Moredsh-agent-enhancer: Enriches Prompts, Fallbacks, Usage Metrics, and Auto Routing in the DSH Desktop Client
dsh-agent-enhancer is a plugin for the DSH Desktop client maintained by rebornace. It extends the official session and settings interfaces with prompt history, optimization, and prediction; restore-to-here and undo-restore; Token billing and context details; automatic model routing; and per-account balance viewing. The article describes host version requirements, installation methods (in-app, npm desktop profile, dsh-market), common shortcut keys, and where to find the relevant settings. It also notes that plugin permissions and privacy behavior should be referred to the README.
Read MoreLaunch Acceptance Officer: Read-Only Check of Launch Status and Evidence-Backed Acceptance Reports
Go-Live Acceptance Officer is an Agent Skill based on the common SKILL.md format, maintained by community publisher user_06b9b4ad. It is categorized under dev-programming and is version 1.0.0. It takes a system address and an optional OpenAPI document as input to perform read-only go-live checks without logging in or modifying production data. It verifies the health of public HTTP/API endpoints and compares a textual requirements checklist against actual interface evidence item by item. Finally, it generates a Markdown-formatted acceptance report with a GO / CONDITIONAL GO / NO-GO conclusion, with each conclusion supported by real interface evidence. It is suitable for pre-release smoke testing, public interface health verification, compiling delivery acceptance reports, and identifying unverified go-live risks. The page indicates `requires_api_key=false`. As no specific installation command is provided in the official materials, it is not fabricated in this article.
Read Moredsh-qwenwork-connector: Connecting Directly to Qwen Office Models in DSH
**dsh-qwenwork-connector is a DSH model inference plugin maintained by REALOIYO that integrates QwenWork into Harness as a provider. It supports device code login, stores credentials in `~/.dsh-qwenwork-connect/accounts.json`, and connects directly to the official API endpoints — no QwenWork desktop client or magpie gateway required. It offers models such as flash, pro, and Qwen3.8-Max with configurable context windows. Installation can be done via `dsh plugin --profile desktop add dsh-qwenwork-connector` (once published on npm) or by installing from a local directory. This is a non-official implementation, requires a purchased subscription, and users are advised to review the source code and MIT license before use.**
Read MoreWorkbuddy-prompt-template: Organizing Office Prompts into Reusable Templates
workbuddy-prompt-template is a community-sourced Agent Skill (@user_7c61b399/workbuddy-prompt-template, version 1.0.0, categorized as ai-agent) positioned as a prompt template library for office scenarios. It provides 17 parameterized prompt templates covering Excel processing, batch file processing, report generation, scheduled tasks, remote control, and content creation, along with a universal prompt framework and 10 advanced tips. The tag requires_api_key=false indicates that it does not require a separate API key. This article describes its positioning, core functions, and general usage flow: users first identify their specific office scenario, then supplement details such as files and output formats, fill the parameters into the template, and submit it to the Agent for execution. The article honestly notes that neither the SkillHub page nor the interface address could be retrieved during verification; therefore, installation commands and official examples are not provided, and users should refer to the official page for activation methods. It also cautions that this is a community-sourced resource with limited scenario coverage, templates require self-validation, and parameter errors may lead to results deviating from expectations.
Read Moredsh-session-trash: Two-stage session deletion in DSH desktop
DSH official persistence only supports archiving and cannot delete session directories. dsh-session-trash provides desktop clients with features to move items to the trash, restore via the settings page, or perform a two-step permanent deletion, while ensuring running sessions are protected. Written in pure JavaScript and licensed under MIT (v0.1.0), installation requires copying the files to the profile's node_modules, registering it in package.json, and then fully restarting DSH.
Read MoreA-Share Theme Supply Chain Penetration: Deconstructing Industry Themes into Hierarchical Supply Chain Maps
**# Preface** When mapping out industrial chains, a common approach is to review several research reports and compile company names into a spreadsheet. While this generates a supplier list, it suffers from three main issues: the list is "flat" and fails to reveal the transmission relationship across upstream tiers; market share figures are static, making it difficult to distinguish between overall industry share and specific supply share within a key business segment; and key players (leaders) and upstream/downstream clients are intermingled, making it hard to instantly identify which entities have the strongest correlation with the investment theme. The Skill "A-Share Thematic Supply Chain Penetration" introduced below replaces manual sorting with a standardized workflow. Starting from end-market demand, it penetrates upstream tiers step-by-step, marking core enterprises, suppliers, and market share at each level to generate a hierarchical supply chain penetration chart. **# What is this?** "A-Share Thematic Supply Chain Penetration" is a research Skill dedicated to industrial and supply chain analysis for the A-share market, with coordinates `@user_98d5
Read Moredsh-minimal-router: Per Stage, Expose Only the Tools Currently Needed by the LLM
dsh-minimal-router (v1.19.5, MIT) is a workflow-class host plugin for DSH. In a single package it offers phase-based tool routing, circuit breaking on consecutive errors, and cache probing. After installation, open Settings → Plugins and toggle Minimal Router off and then on; verify with route_status. According to README tests in a fresh session, the number of tools drops from 165 to 66, and the declarative footprint shrinks by roughly 67%. It does not reduce the system prompt and cannot coexist with another tool router. Runtime data is stored in ~/.dsh/minimal-router/.
Read MoreDSH Session Toolkit: Workspace Migration and Runtime Agent Preset Switching
The DSH client plugin dsh-session-toolkit (v0.1.2, MIT), maintained by QqiqiQ, provides buttons for Workspace migration and Preset switching within the DSH web conversation header. It leverages host routing to accomplish cwd continuation and in-session preset rebinding, features that cannot be implemented solely via the official API. The article details installation commands, the semantics of the migrate/preset interfaces, limitations regarding models and archives, as well as the sources available on SkillHub and GitHub.
Read MoreStock News Radar: Consolidating Scattered Market News into a Holdings Briefing
Stock News Radar is a community-sourced LLM Skill published by user_9456ee1b. Classified under "professional" with version 1.0.0, it requires no API Key. It automatically monitors three levels: domestic and international macroeconomic dynamics, industry sector trends, and messages regarding specific watchlist stocks. It generates pre-market briefings daily and supports manual triggering, aiming to help users identify which news truly impacts their holdings within 30 seconds. The tool supports maintaining watchlists via screenshots or text, features a seven-level impact rating system, and offers news source filtering. It is ideal for developers and analysts who lack time to monitor markets constantly and need to quickly aggregate news relevant to their portfolios before trading. Please note: Users should independently verify the scope of coverage and rating accuracy, as source filtering may overlook niche but valuable information, and assessments of bullish or bearish signals must be made independently based on fundamentals. As this Skill is community-sourced, it is recommended to review its information sources and rating logic before use.
Read MoreBid Document Preparation: Integrating Bidding Document Parsing, Compilation, and Dual-Track Verification into a Single Workflow
Bid document creation is a community Skill based on the SKILL.md format (@user_49227058/bid-studio, category: business-ops, version: 0.1.2, source: community, requires_api_key=false). It consolidates the bidding workflow into a five-step pipeline: mounting industry expansion packs, bid decision-making, tender document parsing, bid compilation (mapping scoring items one-to-one with response sections), dual-track verification (quotation consistency, rejection red lines, word count formulas, and scoring coverage), and delivery packaging (Word output, deviation tables, and mock review reports). The printing service package is pre-configured with scoring paradigms for public security and People's Congress projects, chapter templates for five major components, qualification thresholds, and a risk library. Trigger scenarios cover bid writing, tender parsing, scoring item responses, rejection clause checks, quotation case validation, deviation tables, qualification reviews, commercial and technical bids, printing services, government procurement, framework agreements, and designated printing. The article outlines core features, activation methods, typical use cases, and limitations: dual-track verification relies on explicit terms in the tender documents and cannot replace final human confirmation.
Read MoreNational Fitness Coach Qualification: Sports Nutrition Module - Skill: For Theory Exam Preparation Practice
This article introduces the SkillHub education category Skill "Fitness Coach NVQ · Module: Sports Nutrition" (Version 1.0.2), posted by community author user_af163eaa and sourced from the community. Targeting users preparing for the theoretical sections of the National Vocational Qualification exam, this Skill covers exercises for the sports nutrition module. The questions are AI-generated based on the general syllabus. The official description explicitly states that it is for practice purposes only and does not predict exam questions or serve as an official resource. The tags indicate `requires_api_key=false`. Since the captured data does not provide directly copyable installation commands or configuration fields, this article describes the situation truthfully. It provides the official address, the SkillHub page, the coordinates (`@user_af163eaa/fitcoach-quiz-m04-nutrition`), and the displayed ratings and download figures as verification references. Furthermore, the article outlines the core functions, the access points that need verification before enabling the Skill, the practice path organized according to the data, the target audience, and precautions stating that the Skill cannot replace textbooks or real past exams.
Read MoreDSH-Profile-Vault: DSH Personal Information Vault with Session Authorization
dsh-profile-vault is a DSH community memory plugin that stores personal profiles locally indexed by name. Read operations such as profile_list and profile_get require three-tier authorization from the current session user, with sensitive fields masked by default. It provides five tools including profile_remember, profile_forget, and profile_grants, along with a "Personal Profile" settings page, and supports observation time and asOf queries; v1 is not encrypted. Installation is available via the plugin market or the tgz command from the Release page, and DSH must be restarted after installation.
Read MoreDSH Calculator: Providing Deterministic Mathematical Evaluation for Agents
DSH Calculator (qqiqiq/dsh-calculator) is a DSH plugin maintained by QqiqiQ. It registers a `calculate` tool via a handwritten parser (without eval) to address agents' mental calculation and floating-point drift issues. It supports batch expressions, matrix operations, numerical integration and differentiation, and an exact decimal mode for currency. Installation requires using the official tgz package URL and restarting the dsh web interface. MIT licensed, current package version 1.2.0.
Read Moreedu-physics-3d: Transforming 3D Physics Scenes into Interactive Educational Web Pages
edu-physics-3d is a general-purpose SKILL following the SKILL.md format on SkillHub, published by user_8bb4b4f5 and designed for 3D physics education. It directly generates interactive teaching webpages covering scenarios such as the Lorentz force, atomic orbitals, crystal structures, rigid body rotation, 3D magnetic fields, and electromagnetic waves. The interface features a left column with problem statements, a slider-driven dynamic console, and conservation law indicators; a middle column with step-by-step KaTeX solutions; and a right column featuring a zoomable, rotatable Three.js 3D canvas. Three.js is loaded via CDN, and there is no dependency on SymPy. Structurally parallel to edu-chem-reaction, this skill is exclusively for 3D scenarios; for 2D physics, please use edu-physics. The document provides both Chinese and English trigger words and usage instructions.
Read More