dsh-report-fix: Enforce that task_dispatch rounds do not end before task_report
dsh-report-fix is a DSH workflow plugin maintained by JohnnnSinaaa: a turn initiated by session_call(task_dispatch) must not be closed until a task_report has been successfully sent to the caller. It does not modify dsh-call-session; instead, it tracks outstanding tasks and clears them through the agent/pre-step, tools/post-execute, and agent/turn-stopping hooks, and steers to reopen the turn at turn-stopping. It supports configuration options such as maxNudges and locale, and is installed via `dsh plugin --profile web add <local-path>`. MIT licensed, requires Node >= 20.
Read MoreAdvertising Compliance: Turning Pre-Launch Review into a 4-Step Q&A Process
Ad Compliance (@user_cc440efc/ad-compliance) is an interactive advertising compliance review assistant, with a community source, professional classification, a displayed rating of 100000, 909 downloads, and 0 installs. It collects advertising information through four-step questioning and automatically performs risk identification across eight dimensions (absolute terms, false advertising, prohibited content, comparative advertising, minors protection, pre-approval, statutory labeling, and industry-specific compliance). It leverages the Huayu Yuandian MCP to retrieve laws, regulations, and cases, quantifies penalty risk, and ultimately outputs a structured Word-format review report. Trigger scenarios include advertising review, ad compliance, ad copy review, ad risk screening, advertising law compliance, prohibited-word checks, and pre-launch advertising review. This article explains its positioning, core capabilities, and applicable scenarios, and notes caveats such as no official installation command or verified sample has been released, and that review quality depends on the completeness of questions and the availability of Huayu Yuandian MCP.
Read MoreDX-P5L4-04 GTM Strategy: GTM Strategic Planning Skill for Physicians and Pharmaceutical Companies
An introduction to the DX-P5L4-04 GTM Strategy skill, produced by Baitong AI Service Platform. It is a GTM Strategy segmentation service skill tailored for physician and pharmaceutical company scenarios. Based on the universal SKILL.md format, it operates according to the dual-dimensional physician-pharma matrix of P5 patient medication classification × L4 academic authority positioning, and is triggered when a physician or pharmaceutical company requires a GTM Strategy. The text outlines its positioning, verified functions, skill coordinates and official URL, typical trigger-based use cases, applicable scenarios, and precautions. Regarding the installation command and variations across different editors, the text explicitly states that they could not be verified due to a failed fetch of the official page, maintaining accuracy without speculation.
Read MoreThe AI Copy De-AI Master: Turning AI Text into Human Language
AI Copy De-AI Maestro was released by Guomeng Digital Technology, located at @guomeng/ai-rewriter-master. Current version 1.1.0, it falls under the content-creation category of SkillHub, is sourced from the enterprise sector, and is a paid service that does not require an API key. It only does two things: detect the mechanical and formulaic tone in AI-generated copy, and rewrite expressions into more human-like and warmer phrasing. The official introduction does not specify detection rules, rewrite intensity, or constraints on retaining the original meaning. No reproducible input-output examples or installation commands are provided in the documentation either. Therefore, this article only describes the calling form, while installation and configuration should be subject to the official page. It is suitable for content teams to batch process first drafts but does not have fact-checking capabilities. The rewritten results must be manually verified for dates, numbers, and proper nouns before delivery.
Read MoreHand-drawn Diagrams: Generating Hand-drawn Style Excalidraw JSON Diagrams with Skill
Hand-drawn Diagram is a community-sourced Agent Skill, identified as `@user_db7424f5/udb7-handdrawn-diagram`, published by `user_db7424f5` and currently at version 1.0.0 under the `design-media` category. Its purpose is clearly defined: generating hand-drawn style Excalidraw JSON diagrams, covering three common types—architecture diagrams, flowcharts, and sequence diagrams. It does not require any additional API key at runtime. The output is JSON data that can be read and further edited in Excalidraw, not a static image. It is suitable for developers who need to quickly insert a draft-style diagram into documentation, design reviews, or PR descriptions and still want to adjust nodes and connections later. Specific input parameters, output schemas, official installation commands, and configuration differences across hosting environments have not been verified in the available materials. Therefore, this article only presents the verified information and provides the official page for readers to confirm the latest activation method.
Read MoreAI-Generated Supervision Daily Log Version 2 – Custom Template Support: Auto-Populates Daily On-Site Conditions into User's Own Word Templates
The primary pain point with supervision logs lies in manually entering the daily construction status item by item into the project's existing Word tables. This Skill accepts supervision log templates in any format (.doc / .docx) uploaded by users. The AI first identifies the template structure, then fills it in based on the day's on-site conditions described by the user in natural language, and outputs a fully formatted .docx file. When information is incomplete, the Skill uses an interview-style approach to gather all missing details before generating the document, rather than arbitrarily auto-filling blanks. Based on the standard SKILL.md format, it is sourced from the community, categorized under office-efficiency, and is at version 1.0.0. The publisher is user_0f306ece, and it is officially marked with requires_api_key=false, meaning no API key is needed. The article outlines the verified capabilities, a three-step workflow for a complete usage session, and items that require user verification before use: checking the file processing method due to its community origin, confirming the layout after .doc template conversion, and noting that unverified installation commands and client-specific differences were omitted because the official page's body text could not be scraped.
Read Moredsh-theme-leshan: Three Leshan Photography Skins for the DeepSeek Harness Web Interface
dsh-theme-leshan provides three Leshan photography themes—Sanjiang Night, Lingyun Evening Glow, and Emeimist—for the DeepSeek Harness Web GUI. Each theme derives 119 `--dsw-*` tokens from six input colors and supports both light and dark palettes. Wallpapers cover the body and conversation areas, and can be toggled within the app using the Leshan pill or Alt+K. Installation is available via the official tarball or GitHub source. It is compatible with DSH 0.2.0-rc.1 to 0.2.x and licensed under MIT.
Read MoreCAD Drawing Generation for Hydraulic and Pneumatic System Design
This article introduces the community Skill "Hydraulic and Pneumatic System CAD Drawing Generation" on SkillHub, located at @user_5453610a/hydraulic-system, version 1.0.2, categorized under dev-programming. It encapsulates the generation of CAD drawings for hydraulic and pneumatic systems into a capability that can be triggered by natural language. Trigger phrases include "hydraulic and pneumatic systems," "generate hydraulic and pneumatic system drawings," "create a hydraulic and pneumatic system," and "hydraulic-system." The tag is requires_api_key=false, meaning no additional API key is required. The article starts with the pain point of repetitive drawing among designers, explains the Skill's positioning, verified core functions, trigger methods, and typical usage, and honestly states that official materials do not provide installation commands, directory paths, or configuration file fields, and that information on drawing formats, component libraries, and sheet sizes has not been verified, so no inferences are made. Finally, it outlines the target audience and precautions: the source is community, the installation count is 0, community feedback is minimal, the classification leans towards the development toolchain side, it cannot replace professional CAD software, and safety circuits and pressure checks still require manual verification.
Read MoreUnified Skill Pack for Power Test Development: A One-Stop Agent Skill for Developing Power Test Host Software
This article introduces the community skill “Power Test Dev Suite” on SkillHub. It consolidates end-to-end capabilities for developing power-supply test host software into a single skill package: development workflow orchestration; project-code indexing and hand-off; device manual and command lookup; instrument and device communication abstractions (VISA/CAN/I2C/serial/RS232/RS485/PLC/Modbus/SCPI); test-data and configuration management; logging and security audit; UI generation (PySide6/Qt/WPF); device simulation and automated testing; code review and architecture design. Starting with V1.0.2, it adds project scaffolding, a device-driver registry generator, command-semantics checking, and a zero-dependency smoke-test suite; V1.0.3 further adds data-migration matching strategies, data-extraction and segmentation strategies, a generic external-data file loader, and chart-readability methodologies—all positioned as general methodologies rather than tied to any specific use case. The article describes its layered capabilities, progressive-loading mechanism, trigger keywords, typical usage, applicable scenarios, and limitations (requiring on-hardware validation, protocol adaptation, and version verification), and provides official links and coordinates. It is intended for engineers building power-supply test host applications or designing device-automation test chains.
Read Moredsh-theme-atlas: Six Material Skins, One Token Engine
dsh-theme-atlas is a theming plugin for DSH Web maintained by John-20260101. It features a shared token engine across six material skins, deriving 119 --dsw-* aliases from six input colors per skin, and supports dual light and dark appearances with in-app switching. The runtime is purely CSS-based, containing no bitmaps or host routing. To install, use `dsh plugin --profile web add github:John-20260101/dsh-theme-atlas` or download the official release tarball. It is compatible with DSH Web 0.2.0-rc.1 through 0.2.x and is licensed under MIT.
Read MoreNaming Master Skill Introduction
"Qiming Master" (Name Master) is a "content-creation" Skill published on the SkillHub community, located at @user_6a6f1523/skill001-test. It is version 1.0.0, tagged with requires_api_key=false, has a score of 100000, 957 downloads, and 0 installations. Positioned as an intelligent naming assistant, it is triggered when users need to name or rename individuals (including babies), companies, shops, brands, or online handles, covering four main scenarios: baby naming, company naming, online nicknames, and brand naming. The article outlines its basic information, verified core functions, and official address, while pointing out that the installation command, configuration fields, and typical examples were missing from the retrieved data. As the official page could not be accessed this time, verification is required independently. Additionally, it is noted that since this tool originates from the community, has zero installations, and lacks information regarding duplicate name or trademark checks, it should only be considered a candidate solution prior to formal registration.
Read MoreColor Matching: Design and Aesthetic Skills to Enhance AI's Color Judgment
"Color Palette" (`@user_3c6cb52e/self-ent-design-color-palette`) is a community design and aesthetics skill on SkillHub (Version 2.0.2, Category: design-media, Source: community). Centered on color schemes with the trigger word "color-palette," it features a senior visual designer persona and requires no API Key. This article outlines the problems it solves, verifiable capability boundaries, the verification status of its installation and activation methods, its target audience, and key considerations.
Read MoreDong Xiaoyu Compliance Consulting Skills: Listed Company Compliance, Disclosure, and Qualification Verification
"Dong Xiaoyu Compliance Consulting skills" is an Agent Skill (version 3.0.1, source: community, category: professional) maintained by community publisher user_160536ff. It is used for consulting on listed company compliance, appointment qualifications, corporate governance, information disclosure, announcement formats, and references to similar announcements. The workflow is as follows: first, the hegui MCP is called to query and verify the current full text in the regulations database; the structured conclusions from the regulations are used only for internal routing. Only when the regulations support a disclosure obligation are cases of similar announcements retrieved. When the provisions indicate that an announcement must be issued for a particular format or matter, the current announcement format documents and template attachments for the corresponding board are additionally retrieved. The final answer must include the official regulation name, document number, status, original text of the provisions, and applicability analysis, and must be accompanied by the announcement format templates and the original files of announcement cases that have passed verification. Independent web searches or supplementing regulations and announcements from model memory are prohibited. No installation command is provided in the official materials, and the tag is requires_api_key=false.
Read MorePaperConan: A Numerical Health Check for Paper Source Data Tables
PaperConan is an open-source command-line tool and AI skill designed to perform numerical health checks on source data tables within paper supplementary materials. It utilizes over 30 deterministic detectors to inspect tables in .xlsx, .csv, .tsv, PDF, and Word formats, identifying signals such as constant column ratios, cross-table duplication, row-wise reuse, shared decimal tails, GRIM/GRIMMER mean consistency, and abnormal last-digit distributions. Findings are pinpointed precisely to the specific file, sheet, row, column, and rule. Scans generate `audit/scan.json` and `report.html`, and the tool allows retrieval of public supplementary data by DOI via `paperconan fetch`. All results represent statistical signals rather than definitive conclusions and require manual verification against raw data, figure captions, and Methods sections; the review of image assets is delegated to external multimodal agents.
Read Moredsh-seesion-copy-enhance: Copy entire messages (text and attachments) in DSH
The built-in DSH copy function outputs plain text only; messages containing images or files cannot be fully transferred across sessions. dsh-seesion-copy-enhance (JMILI, v0.2.0, a client plugin) takes over message copying. Through Host capture/resolve and clipboard text/plain, text/html, it restores actual composer draft attachments upon pasting. It supports DSH 0.2.x and can be installed via the plugin marketplace or `dsh plugin add`. Optional size and history configurations are available.
Read More**Loan/Mortgage Calculator Skill: Accurately Calculate Monthly Payments and Total Interest**
This is a community-sourced model skill located at @indiv-simon/dw-loan-calculator. Version 1.0.0. Category: professional. No API key required. It is triggered when users mention terms such as mortgage calculator, monthly payment, loan interest, prepayment, equal principal and interest, or equal principal. Input the loan principal, annual interest rate, and term to calculate the monthly payment and total interest, with support for comparing the two repayment methods: equal principal and interest and equal principal. The article outlines its capability boundaries, integration principles, and typical use cases. It also points out pre-integration items that require actual testing for verification, including the prepayment calculation logic, an install count of zero, and its community-sourced nature.
Read Morelove-suite: DSH Local-first Companionship and Memory Plugin
love-suite-public (dsh-memory-network) is a DSH companion-type memory plugin maintained by jli517166-gif. Version 1.16.0 is currently in Early Testing. Memory writes target local files in the love workspace and support persona switching, core memories and a reflection queue, keyword search with optional semantic retrieval, a mood heartbeat, proactive chat, an admin dashboard, and a memory constellation map. QQ/group chat channels are optional. Install with: dsh plugin add github:jli517166-gif/love-suite-public. It is not intended as your sole memory store; before using, back up memory/ and be aware of the limitations of a non-stable release.
Read MoreJavaScript: Comprehensive Style Guide Skill
This article introduces JavaScript Skill (@zcwl/javascript-skills, version 1.0.2). It is a comprehensive JavaScript style guide designed for development and programming scenarios. Once activated, it serves as a best-practice reference during coding and review phases, with the tag requires_api_key=false. The article describes the Skill’s positioning, core capabilities, official coordinates and web entry point, and general activation workflow, as well as applicable scenarios and limitations: it does not replace ESLint, TypeScript, testing, or build tools, nor does it claim any runtime capabilities. Where specific installation commands and client-specific integration differences could not be verified, the article notes that the official page should be consulted.
Read MoreThe R&D Efficiency Performance Instrument: Transforming Leadership Subjective Assessments into Calibrated Scoring
The Dev-Performance-Evaluator is a data analysis skill released by the community, located at @user_fa14b5bc/dev-performance-evaluator. Version 1.0.0, it is categorized under data analysis. Targeted at Technical Leaders and HRBPs, it accepts input data on code commits, requirement delivery, defects, and collaboration. It outputs individual and team efficiency scores across four dimensions: delivery efficiency, delivery quality, technical contribution, and collaboration impact, while providing ranked comparisons stratified by job level. It features a built-in three-level context calibration: programming language coefficient × project complexity coefficient × legacy system discount × new hire protection period, designed to mitigate unfairness caused by language, complexity, and technical debt. The official documentation also lists five anti-gaming rules: requirement splitting detection, formatted commit exclusion, LGTM quality discounting, comment ratio filtering, and down-weighting of pure code-less reviews. The official page does not provide direct installation commands or configuration fields for verification, so this article makes no assumptions regarding these details. When processing code commits, defects, and collaboration data, it remains necessary to confirm data scope and permissions. Official data indicates requires_api_key=false, with scores of 100000, downloads at 940, and installs at 0; this suggests limited public usage experience, and small-scale retrospective validation is recommended prior to implementation.
Read MoreVideo to Text: Converting Local Videos or Public Links into Clean Transcripts
"Video-to-Text" is a community-sourced Agent Skill (located at @user_b9c6802d/guaikei-videotoext), currently at Version 1.0.0 and categorized under data-analysis. It converts local video files or public video links from platforms like Douyin and Xiaohongshu into clean transcripts with redundancies removed. It supports customizing derivative content based on user-defined prompts, such as general summaries, social media copy (Xiaohongshu/Douyin/WeChat Official Accounts), quote extraction, meeting minutes, course structuring, livestream reviews, interview transcription, and Chinese-English translation. This skill is ideal for video-to-text conversion, subtitle extraction, speech-to-text, video summarization, copywriting extraction, video translation, or short-video secondary creation. It is not applicable for livestream streams, links requiring login/membership/encryption, purely instrumental music videos, or batch processing. An API Key is required for use.
Read MoreDisk Division Steward: Read-only Multi-disk Space Panel under DSH
The DSH client plugin **dsh-drive-manager** for the JKL-3484, designed for Windows 10/11, automatically detects fixed drives (e.g., C/D/E/F) and employs a green, yellow, and red color-coding system to identify cleanable items, user data, and system files. It analyzes download directories from browsers and applications such as WeChat, providing recommendations for multi-drive allocation. Featuring a read-only design, the plugin performs no deletions. Installation requires DSH version 0.2.1-rc.2 or higher, using the `dsh plugin add` command from the `desktop` profile. After exiting the system tray and restarting, it becomes accessible from the sidebar.
Read MoreMind Map Generation Pro: Adding a Set of Verifiable Constraints to AI Mind Map Generation
Directly asking models to generate mind maps often leads to issues such as overloaded primary branches, meaningless leaf nodes, random coloring and layout, and poor readability after export. Mind Map Generation Pro (@kunlungrowth/kunlun-mindmap-pro, v1.1.0, source: enterprise, publisher: Kunlun Growth, category: content-creation, no API Key required) employs a set of rules to constrain the generation process: selecting schemas based on usage, converging central themes, setting limits on primary branches, ensuring leaf nodes are actionable, standardizing color and layout, and validating export readability. It also includes a comparison guide for five common diagram types, thresholds for identifying overload, and root cause analysis for rework. This article summarizes the feature list, installation entry points, and a typical six-step workflow based on the official introduction, while noting limitations such as the absence of specific installation commands and numerical thresholds in the provided documentation.
Read MoreIn-Depth Analysis of Corporate Equity: Turning Equity Penetration into a Deliverable Report
Introducing the SkillHub community skill “Enterprise Equity In-Depth Analysis Report,” maintained by user_87fe9f99, version 1.0.2, categorized as professional, tagged requires_api_key=true, and dependent on the Shuidi Credit MCP. It turns equity due diligence into a pipeline: call the Shuidi Credit MCP to fetch data, then generate an in-depth analysis report on shareholding structure and affiliated companies, with outputs produced in sequence as HTML, SVG+PNG, and PDF. The report always covers five modules: ultimate controller penetration, four-layer visualization of the equity structure, a six-tier assessment of control strength, a panoramic mapping of affiliated entities, and disclosure of key risks. Use cases include verifying ultimate controllers of Hong Kong/mainland listed companies, mapping equity structures of unlisted enterprises, and multi-layer penetration through BVI/trust structures. In clients that support the general SKILL.md format, load it via the reference @user_87fe9f99/shuidi-equity-report; an API key must be configured before enabling it. Specific installation commands and configuration fields are not provided; refer to the official page for details.
Read Moredsh-context-manager: Per-Round Context Removal and Session Continuation Prompt
dsh-context-manager is a DSH memory plugin maintained by JKBJ828, which provides context and prompt management at the bottom of a session: through surface replacement, specified turns are masked to prevent the model from seeing them, while original logs are retained; users can view the complete system prompt and edit the supplementary section of the current session, with prompts from other plugins being read-only. It supports the /context command and a web GUI, requiring DSH 0.2.1-rc.3 and session v4. Installation involves cloning the repository, running install_bundle, and restarting DSH.
Read Moreawesome-ChatGPT-repositories: Delegating ChatGPT Repository Curation to Agents
The `awesome-ChatGPT-repositories` is a community-released Skill categorized under `ai-agent`, located at `@user_3c6cb52e/awesome-chatgpt-repositories`. It is Version 1.0.0 and is tagged with `requires_api_key=false`, indicating that no API key is required. Its official positioning is a dedicated skill collection designed to assist AI Agents in efficiently handling the organization and aggregation of ChatGPT-related repositories. The available data provides only the positioning, coordinates, version, source, and classification; it does not expose the SKILL.md body, installation commands, configuration fields, or reproducible examples. Therefore, this article focuses solely on verifiable content, treating the crawled snapshot as the standard for conflicting data such as ratings or download counts. Applicable scenarios include repository directory maintenance, preliminary screening of Awesome lists, link availability checks, and topic classification assistance. Please note that it originates from the community, and do not automatically attribute network scraping, code execution, or automatic commit capabilities to it.
Read More