Customer Requirements Confirmation Skill: Finalize Delivery Scope and Acceptance Criteria Item by Item Before Kick-off
The "Customer Requirements Confirmation Skill" (community, v1.0.1, business-ops) extracts requirements from customer chat logs, organizes them into a checklist for item-by-item confirmation, and generates professional confirmation scripts to help novice freelancers define the delivery scope, acceptance criteria, and project boundaries before starting work. The article introduces its three core capabilities, the SkillHub access point, and a typical three-step usage flow (chat logs → requirements checklist → confirmation scripts). It also notes that the skill is community-sourced, has 0 installs, lacks large-scale validation, and that key items require manual review.
Read Moredsh-maa-cli: Driving maa-cli with Natural Language in DeepSeek Harness
dsh-maa-cli is a DSH workflow plugin released by maintainer nhyo, enabling control of maa-cli through natural language and the /maa local command, with a bundled standalone runtime. It supports plan confirmation, roguelike strategy compilation, Box template copying, leveling, and goal-driven workflows. To install, follow the README: dsh plugin --profile desktop|web add dsh-maa-cli; the initial download is approximately 520–630 MB for the runtime. The plugin is licensed under MIT and requires Node >= 22.
Read MoreMaster of Tendering Document Element Extraction: Extracting Structured Key Information from Bidding Documents
"Tender Document Element Extraction Master" is a general-purpose Skill in SKILL.md format maintained by user_e2cb3698, a publisher in the SkillHub community. Located at @user_e2cb3698/bixg-bs, it is marked as version 1.0.2 and sourced from the "community" category under "business-ops." The tags indicate `requires_api_key=false`, meaning no separate API Key configuration is required. It is designed to extract structured key information from bidding or tender documents based on fixed dimensions, covering basic project information, mandatory content, key details, scoring mechanisms, and detailed rules. Listed trigger scenarios include "extract bidding document info," "analyze bidding documents," "read tender documents," "key document info," "extract mandatory bidding info," "generate bid Excel," and "bidding scoring table," making it suitable for on-demand invocation within the Skill Editor. When using this tool, it is recommended to focus on one specific requirement at a time. Scanned copies or image-based PDFs require document parsing first, and key content such as scoring clauses should be cross-checked against the original text. As the attempt to scrape the official page was unsuccessful, this text does not provide installation commands or configuration fields that have not been confirmed by the official page, nor does it include unverifiable scoring or download data.
Read Moredsh-zpdf: PDF Parsing and Translation Tool Calling ZPDF in DeepSeek Harness
@nguyenvanhoang1/dsh-zpdf provides tools such as zpdf_parse_pdf, zpdf_translate_pdf, and zpdf_convert_markdown for DeepSeek Harness. It integrates with the Zhiyi PDF API and supports persistent configuration of ZPDF_API_KEY via both Web settings pages and the CLI. Based on the official README, this article covers installation commands, the capabilities of the six tools, optional composition parameters, and three API key configuration methods. It is intended for DSH developers holding ZPDF Plus or Pro accounts.
Read Moredsh-subagent-acp-ui: Integrating Multiple ACP Sub-Agents into DSH via Configuration Page
dsh-subagent-acp-ui is a DSH workflow plugin maintained by NewSpringWei. Based on the official @deepseek-ai/dsh-subagent-acp, it provides configuration pages, handshake detection, and the acp-subagent-routing skill for multiple ACP CLIs; it does not implement the ACP client itself. The recommended installation source is github:NewSpringWei/dsh-subagent-acp-ui#v0.1.0, managed under "Plugins → ACP Sub-agents." For each target, it generates an acp_<target> provider, a subagent_acp_<target> tool, and a /acp-<target> command. It has been verified with six agents, including OpenCode and Kimi Code, and the compatibility list can be updated online. Please note the official limitations, such as the model being fixed per target, one-time delegation, and the absence of a process pool. Licensed under MIT; requires DSH >=0.2.0-rc.2.
Read MoreAnki Deck Generator - Shadow Math: Converting Materials into Importable .apkg Decks
Anki Card Generator - Shadow Math is a community Skill on SkillHub (Coordinates: @user_63b2cc71/anki-card-generator, Publisher: user_63b2cc71, Version: 1.0.1, Tag: requires_api_key=false). It organizes materials such as images, Excel, Word, PDF, and TXT/Markdown into Anki cards and packages them into .apkg files ready for direct import into Anki. It features five built-in card models: Basic Q&A, Word/Term, Cloze (Fill-in-the-blank), Formula, and Image. The Word/Term model includes two templates: "Word" and "Sentence Memorization," automatically generating both word cards and sentence cards from the same note data. All templates are optimized for AnkiDroid and AnkiMobile. Word cards are interactive (tapping a word reveals the definition and replays the audio; tapping an example sentence reveals the Chinese translation and replays the audio), while sentence cards automatically display the Chinese translation after the audio reading finishes. Dependencies, including genanki and the Anki desktop application, will be installed automatically. It is suitable for teachers, students, and parents to create memory cards, memorize words and sentences, and review knowledge points.
Read MoreChatbot Assistant: Structuring Business Automation of Repetitive Tasks
"Chatbot Assistant" is a Skill released by Zhiyu Technology on SkillHub, identified as `@org-02qudk26/chatbotkit-automation` (Version 1.0.0). Sourced from the "enterprise" category and classified as an "ai-agent," it requires no API Key. Targeting white-collar professionals, it handles repetitive tasks related to business automation and rapidly organizes results into structured, reusable work deliverables. Since official documentation only provides a brief introduction without reproducible examples, interface names, or configuration fields, this article focuses solely on its positioning, core functions, installation precautions, and applicable scenarios. It does not construct examples or speculate on loading differences across various hosts.
Read Moredsh-gfm-alerts: Render GitHub Alert Blocks as Callouts in DSH
dsh-gfm-alerts is a DSH Web client plugin maintained by NEVSTOP-LAB. It renders > [!NOTE], > [!TIP], > [!IMPORTANT], > [!WARNING], and > [!CAUTION] in replies as GitHub-style rounded callout boxes, supporting Chinese and English titles with light/dark themes. Installation requires dsh >= 0.2.0-rc.1; run the official command dsh plugin --profile web add github:NEVSTOP-LAB/dsh-gfm-alerts, then restart and refresh for it to take effect. It only modifies the interface display, without changing conversations or copying original text.
Read Moredsh-deepsearch: Multi-turn In-depth Research within the Same Session with Disk-persisted Citation Reports
An introduction to the DSH plugin dsh-deepsearch (v1.1.45) maintained by NeptuneRY: it features multi-round query decomposition, parallel sub-agent retrieval and summarization within a single session, generating cited Markdown reports that are saved to disk in every round. This section compares its differences with Agent Search, details the panel and deepsearch tool entry points, installation command (`dsh plugin add dsh-deepsearch`), dependencies on DSH >=0.2.1-rc.1 and Better Sidebar, as well as applicable scenarios and permission considerations.
Read MoreBusiness Plan (BP) Generator: Generating Investor-Oriented Roadshow BPs via Structured Q&A
The "Business Plan BP Generator" is an enterprise-source Skill in the business-ops category on SkillHub, located at @org-vc3t81nj/financing-bp-builder. It is maintained by Bowei Management Consulting and tagged requires_api_key=false. Designed for companies seeking equity financing, it captures information—including business, market, product, growth, team, financials, and use of funds—through a Q&A-based, stage-by-stage co-creation process, and then structures it into an investor-facing PPT pitch deck based on Sequoia / YC methodology. The article covers the core features, installation and configuration items to verify before enabling it, a typical workflow that progresses through phased Q&A, as well as applicable scenarios and caveats: market sizing, financial projections, and financing amounts must be provided by the company with real supporting evidence; the Skill is responsible only for structured generation and cannot replace business judgment.
Read Moredsh-unified-desktop: Unifying Two Windows Desktop MCPs into a Single Interface with 29 Tools
When both dsh-computer-use-win and Windows-MCP are loaded simultaneously on Windows, the model sees 44 tools, duplicate capabilities, and exposed high-risk tools. dsh-unified-desktop merges these into a single MCP based on a capability ownership map, exposing 29 mcp__desk__* tools with built-in policy guards. Based on the official README, this article covers the division of responsibilities, installation prerequisites, the patch-backend patch, disabling the standalone wincu, and troubleshooting essentials. It is intended for developers already doing desktop automation with DSH.
Read MoreWind AI - Stock Trading Strategy: A Skill for Generating Structured Stock Trading Strategy Reports
This article introduces the Wind AI Stock Trading Strategy Skill. Published by Wind Information and categorized as "professional" (version 1.1.2), it requires an API Key to run. Users input a stock code or name along with the holding cost price, and the Skill generates a structured trading strategy report in Chinese Markdown. The report covers eight major sections: position profit and loss, fundamental valuation, technical quantitative indicators, industry ecosystem, news events, execution strategy, and risk assessment. It is suitable for scenarios including individual stock trading strategies, position analysis, buy/sell decision references, support/resistance level analysis, and comprehensive analysis of technical, fundamental, and news factors.
Read MoreWriting Style Skills: Establishing Writing Standards for Agents via Community Skill
The Writing Style Skill (`@user_3c6cb52e/writing-style-skill`) is a community-published AI-agent skill. Currently at version 1.0.1, published by `user_3c6cb52e` and tagged with a community source, it holds a rating of 100,000, 1,177 downloads, and 0 installations. It addresses the pain point of inconsistent writing styles across Agent outputs by extracting scattered prompt requirements for writing style and encapsulating them into a Skill, ensuring that Agents execute writing tasks according to a unified set of rules. The tag `requires_api_key=false` indicates that no additional API key is needed. The article explains its positioning and the problem it solves, noting that the official description only states "helps AI Agents efficiently complete relevant tasks" without listing the supported writing styles, specific rule items, or injection methods, requiring users to verify these details directly on the page. Since the retrieved data does not include installation commands, directory conventions, or configuration fields, the article refrains from providing speculative commands. Instead, it outlines the general approach: first open the SkillHub page to install via its prompts, then integrate the Skill into the Agent environment (noting that paths differ for Cursor, Codex CLI, and Claude Code). The usage section provides a general framework rather than officially runnable examples: issue a writing task and declare adherence to the Skill's rules, confirm the Skill is loaded, and review the initial draft for style before checking facts and formatting. Finally, it highlights applicable scenarios and potential pitfalls: unclear functional boundaries, the need for independent verification due to its community source, a lack of case studies due to 0 installations, and the fact that stylistic consistency does not equate to content correctness. The two entry points are the official address and the SkillHub page; both failed to be retrieved during this writing session, so page details should be based on actual access.
Read MoreExpert in Designing Student Learning Outcome Evaluation and Feedback Rubrics: Implemented in Word
This community Skill (located at @user_72b898e0/rubric-design-expert, version 2.1.12, category: education, requires_api_key=false) is designed for K-12 and higher-education teachers as well as trainers. It integrates the creation of Student Learning Outcomes Rubrics and the generation of Word documents into a single workflow: users need only specify the assignment type (e.g., homework, exam, project, paper, presentation, lab, report) and the assessment dimensions; the Skill organizes these into a structured rubric and outputs a ready-to-distribute Word file. Triggering is based on intent rather than a fixed keyword list. Common trigger phrases include 作业评价、学生学习效果评价、评价量规、评分量规、评价量表、评分量表; intents such as Rubric, grading rubric, 评分细则, or the colloquial “How to set a scoring standard for student assignments” are treated as equivalent. The article explains the problems it addresses (eliminating manual formatting and ensuring consistency of standards across instructors), verifies its core functionalities, provides coordinate and configuration details, presents reproducible typical use cases, and outlines key notes: it generates a rubric document rather than an automatic grading system; more specific inputs yield better outputs; subject-specific standards should be reviewed by the course instructor; installation commands and supported editors vary by environment and cannot be independently verified in public sources—refer to the official page for the latest information.
Read MoreSoft Exam Thesis Framework Generator: Generate Thesis Outlines Based on Topics and Project Backgrounds
In the preparation for the Soft Exam senior level, the difficulty of the essay section lies in how to structure a personal project background into an outline that graders can easily recognize once the topic is given. Targeted at examinees preparing for senior-level qualifications such as the Information System Project Manager, System Architect, and System Planning and Management Professional, this community-sourced Agent Skill (ID: @user_7b66698d/ruankao-essay-generator, Version 1.0.0, Category: Education) automatically generates essay outlines that comply with official scoring standards based on the user's inputted exam topic and personal project background. Trigger scenarios include Soft Exam essays, senior-level Soft Exam essays, Information System Project Manager essays, System Architect essays, and Soft Exam essay outlines. Data shows that requires_api_key=false; as no specific installation commands or configuration fields are provided, this information is stated as-is in the text, and readers should refer to the official address for the latest updates. To use it, input the exam topic and project background together, first verify that the generated outline covers all topic requirements, and then proceed to expand it into a full-length essay.
Read Moredsh-sidebar-expand-all: Expand all sidebar sessions at once, collapse strictly back to 5
By default, the DSH desktop sidebar shows 5 more conversations at a time. The community plugin `dsh-sidebar-expand-all` (mzjqwq, MIT) replaces two parts of the watermark logic in `client.js` within a same-size patch package: expanding once displays all remaining conversations, while collapsing is fixed to 5. It supports installation via `dsh plugin --profile desktop`, a local web page for Apply/Restore, and CLI scripts; DSH must be fully restarted after modifying the `.asar` archive. The plugin includes backup and integrity checks, and the patch must be reapplied after upgrading DSH.
Read MoreiOS Development Expert: Embedding Swift/SwiftUI Native iOS Development Requirements into Skill
The iOS Development Expert is a general-purpose SKILL.md format Skill released by Zhiyu Technology and classified under dev-programming, located at @org-02qudk26/ios-developer, version 1.0.0, with enterprise source marking and no API Key required. It solidifies native iOS application development requirements into reusable instructions, with capabilities covering Swift/SwiftUI native development, iOS 18 version baseline, UIKit integration, Core Data, network programming, and App Store-related content (the official description is truncated here). The article explains its positioning, core capabilities, installation and activation methods, limitations of typical usage, applicable scenarios, and precautions. It should be noted that the official documentation does not provide specific installation commands and reproducible call examples. The metadata shows 1183 downloads, 0 installations, and a rating of 100000, with a very small sample size, which is faithfully annotated in the text without speculative completion.
Read Moredsh-local-delegate: Delegating Mechanical Text Work to Local Small Models
The DSH plugin **dsh-local-delegate** (v0.5.1) delegates mechanical text tasks—such as translation, summarization, extraction, and classification—to Ollama or OpenAI-compatible local endpoints. The plugin autonomously reads files, performs chunking, and writes results back, ensuring that long documents do not enter the main model's context. It supports GUI configuration, configurable local AI intervention levels per workspace, and automatic discovery of `local_worker` and `SlaveAgentTeams` for parallel processing. Installation requires downloading the Release according to the official DEPLOY instructions and deploying it via `plugin_manager install_bundle`.
Read MoreCreative Director Pro: A Skill for Delegating Marketing and Creative Management Workflows to Agents
Introducing the Community Skill “Creative Director Pro” on SkillHub: Published by u_8ee34d77, version 1.0.1, categorized under design-media. Targeted at enterprise teams and professional organizations, it covers marketing promotions, advertising placement, customer acquisition & conversion, and growth/viral expansion scenarios. Core capabilities include a creative management workflow, structured outputs with error handling, ready-to-use results, native Chinese interaction with zero configuration, and no API key required. The article describes its positioning, capabilities, how to reference coordinates, typical usage patterns, applicable scenarios, and notes (e.g., the core capabilities section in the official description is incomplete, zero installs, and strictly for legitimate marketing use only).
Read Moredsh-skills-symlink: Load Skill Root Directory by Project
dsh-skills-symlink provides project-specific Skill root directory configuration for DeepSeek Harness 0.2.x: it declares skillPaths via .dsh/skill-paths.yaml and, together with the plugin's maxDepth setting, scans the category/skill/SKILL.md layout, with optional allowExternalPaths to reference paths outside the repository. It also introduces the installation command for v0.4.0, the cordis.patch.yml configuration, typical YAML examples, and the post-install-check self-test, and explains the differences from the Hermes skill symlink concept and customSkillDirs.
Read MoreMiddle School Biology Instructional Design Assistant: A Tool for Generating and Refining Instructional Designs
"Junior High School Biology Course Design Assistant" is an education-category Agent Skill on SkillHub, maintained by community publisher user_cce12b87. Version 1.0.0, no API key required. It helps junior high school biology teachers design and optimize instructional designs and supports three modes: lesson-preparation assistance (generating a complete instructional design from scratch), diagnostic optimization (analyzing an existing draft and iteratively improving it), and key-component design (deep-dive design for seven components: teaching objectives, learner analysis, situational introduction, teaching activities, experimental inquiry, homework design, and assessment design). Through guided conversations, the Skill gradually collects topic information as well as learner and teaching contexts, intelligently recommends the most suitable instructional frameworks, and incorporates multiple instructional design frameworks and strategies. It is based on the 2022 edition of the Compulsory Education Biology Curriculum Standards, with students at the center and teachers playing a guiding role. It is triggered when users mention keywords such as instructional design, lesson preparation, biology instructional design, junior high school biology lesson preparation, biology classroom, instructional design assistant, or express the corresponding intent. This article presents its positioning, core features, installation access, and typical usage.
Read MoreSpreadsheet-to-Lightweight System Generator: Transforming Shop Floor Paper Forms and Excel into LAN-Based Lightweight Business Systems
Addressing scenarios in small and medium-sized manufacturing workshops that are often not covered by ERP/MES systems—such as equipment ledgers, maintenance records, piece-rate labor, patrols, inspections, materials, and supplier directories—this Skill transforms inputs like tables, Excel files, and paper slips into a lightweight electronic business system. It does not replace existing systems but serves as a supplementary layer. Its capabilities include data entry, querying, filtering, Excel export, and statistical dashboards via browsers or mobile phones within a Local Area Network (LAN); multi-user collaboration with role-based access control; master data governance; and read-only AI data access via MCP. No coding is required. Coordinate: @indiv-workbuddy/sheet-to-system. Version: 2.0.0. Source: Community. Category: business-ops. Requires API Key: False. SkillHub Metrics: Rating 100,000, Downloads 1,101, Installations 0. This article outlines the positioning, core capabilities, typical usage paths, applicable scenarios, and limitations. Please note: Specific installation commands, directory locations, configuration fields, and generation parameters could not be verified this time due to a failure to retrieve the official page. Before implementation, please refer to the official address page for instructions. Do not blindly copy installation commands from other Skills.
Read MoreLiznr Sales Leads and Customer Operations | Jianshi AI: Introduction to the Skill for Synchronizing Meeting Notes and Action Items
This article introduces the community Skill "Liznr Sales Leads & Customer Ops | Jianshi AI", published on SkillHub by @user_fccf428b (category: business-ops, version 1.0.0, no API key required). The Skill targets the Liznr AI meeting assistant, covering the full workflow for the Chrome extension and Teams app—meeting recording, real-time transcription, context intelligence, and synchronization of action items to Jira/Slack/Notion—and also provides platform troubleshooting, comparisons for choosing between Fathom/Fireflies/tl;dv, evaluations for recruiting and legal scenarios, multilingual accuracy, and privacy/security Q&A. It lists the official access link, how to enable it via SKILL.md, six typical use cases and applicable boundaries, and reminds readers to verify the community source and platform permissions.
Read Moredsh-session-superagent: Session-level supervisor that wakes up DSH upon repository activity
dsh-session-superagent is a DSH workflow plugin maintained by MYCF711 (version 0.9.166, MIT license). It polls GitHub issues/PRs within a session, obtains credentials via the gh CLI with unified redaction, wakes the current session using Agent.followup when changes occur, and processes them according to the repository's AGENTS.md. It supports watch_start/stop/ack, filtering, rate limiting, activeHours, and isolated worktree. Requires DSH ≥ 0.2.1-rc.5 and a logged-in gh; install with `dsh plugin install dsh-session-superagent` into your profile.
Read Moredsh-cookie-guard: Weekly health checks for DSH login cookies to prevent request headers from causing white screens
The DSH desktop client generates new `dsh-auth-*` login cookies upon every startup. Long-term accumulation of these cookies can exhaust the approximate 16 KB soft limit for request headers, resulting in a blank screen. `dsh-cookie-guard` (mxw0808) addresses this on Windows by monitoring the cookie store via a scheduled task every Monday. If the default count exceeds 40, the plugin backs up the entire cookie database before deleting `dsh-auth-*` entries. Enabling or disabling the plugin within the application controls whether the task executes. To install, use `dsh plugin --profile desktop add` after cloning or extracting a ZIP archive. Manual execution of `guard.py` is also available for maintenance checks. Windows only; requires Python 3.8+. MIT License.
Read More