Seedance2-Storyboard-Generator: Converting Stories or Articles into Seedance 2.0 Storyboard Scripts
This article presents Seedance2-Storyboard-Generator, an Agent Skill designed for the Seedance 2.0 platform. After inputting a story or article, it produces production-ready text specifications through a five-step workflow: analyzing the input, confirming production parameters like visual style and aspect ratio, generating a 20-episode screenplay structured in four acts, creating character, scene, and prop material prompts numbered as C/S/P, and outputting per-episode Seedance storyboard scripts with timelines (including material upload lists, timeline prompts, and end-frame descriptions). It finally delivers three Markdown files: the screenplay, material list, and storyboard. The article provides the ClawHub installation command, timeline prompt templates, a keyword library for camera movement, deliverable file formats, as well as four quality checks (cross-episode continuity, sensitive words, prompt length, and style consistency) and common pitfalls.
Read MoreReal Estate Guardian: Clearly Defined Processes and Pitfalls for Buying, Selling, and Upgrading Homes
The "Real Estate Transaction Escrow Master" is a general-purpose, SKILL.md-type Skill that provides checklists and risk warnings for the entire process across three distinct lines: buying, selling, and exchanging properties. Published by the community account user_af163eaa at the coordinate @user_af163eaa/z-life-house, it is version 1.0.0, categorized under life-service, and sourced from the community. Tagged with requires_api_key=false, it does not rely on external API keys or connect to real estate data interfaces. It covers nodes such as property viewing and inspection, loan schemes, contract signing, title transfer, taxes (deed tax, VAT, individual income tax), and final handover, while identifying common risks within the workflow. It explicitly states that local policies and formal contracts take precedence and does not act as an agent for transaction procedures, preventing Agents from making excessive commitments under the guise of consultation. Regarding installation, official documentation lacks specific commands or differentiated configurations for Cursor, Codex CLI, or Claude Code; only the coordinate can be referenced, and no API key is required. Triggers rely on natural language phrases such as buying a house, selling a house, second-hand housing, real estate transactions, mortgages, title transfer, buying process, selling precautions, second-hand pitfalls, property transfer, how to buy a house, I want to buy a house, what to note when selling, or how to trade second-hand homes. It is suitable for scenarios requiring complete checklists for validation, identifying process omissions, or listing risk points in advance. Limitations include the fact that policies vary by city and time, and the checklist cannot replace contract review, loan approval, or property rights verification. It is a full-text task list Skill.
Read MoreDSH-always-there: Empowering DSH agents to speak for themselves
dsh-always-there is a DSH workflow plugin maintained by LingDangZaiXiang that wakes up agents in the current session via events (log markers) or "gentle" heartbeats. It allows the model to decide whether to speak or return `SILENT`, rather than acting as a scheduled broadcast or relaying messages from a hidden session. The plugin is disabled by default after installation; to enable it, you must configure it in `cordis.patch.yml` and provide a `sessionId`. When integrating with chat software, the `SILENT` output must be filtered. The README indicates that the solution is incomplete and users must accept associated trade-offs, such as visible wake-up prompts. As the plugin runs with the permissions of the DSH process, the source code and MIT license should be reviewed before installation.
Read Moredsh-idea-box: Turn Your Inspirations into Markdown with Enter Key in DSH Web Interface
dsh-idea-box is a DSH plugin maintained by LINFEICXY that provides a round light-bulb button in the Web UI: type a line or drag in/paste a screenshot, then press Enter to save into an .md file inside the “灵感” (Inspiration) folder under your working directory. Screenshots are stored in an attachments subfolder and referenced via relative links. It uses no standalone database; absolute paths are configurable, with settings stored in DSH_HOME/dsh-idea-box.json. Limits: 2000 characters per note, up to 6 images per note, and 8 MB per image. Installation: dsh plugin add dsh-idea-box. Licensed under MIT.
Read MoreIntroduction to the Multi-channel Copywriting Skill: Community, Official Accounts, Video Channels, Images, Bilibili, etc.
This article introduces a multi-channel operational copywriting generation workflow Skill (@user_ff2c8d0a/multi-channel-copywriting, version 1.0.0, category: content-creation, source: community, requires_api_key=false). It processes input materials such as articles, livestream information, product descriptions, or event details to generate ten categories of copy in a single run: WeChat Official Account titles (primary and secondary), WeChat Video Account bios, community forwarding copy, "Little Green Book" (Xiaolvshu) posts, WeChat Official Account cover titles, livestream cover posters, livestream opening remarks, livestream closing remarks, livestream announcements, and livestream community forwarding copy. Trigger phrases include: copywriting, title, bio, community forwarding, cover image, poster, opening remarks, closing remarks, announcement, Little Green Book, four channels, and multi-channel draft generation. Installation commands and cross-platform variations were not provided in the source materials, hence omitted. As of the data extraction date, the SkillHub page displays a score of 100,000, 929 downloads, and 0 installations.
Read MoreK12 Intelligent Question Bank Tool: Organizing Zip-format Question Databases into a Traceable Material Library for Exam Paper Assembly
An Agent Skill designed for K-12 question resource management. It parses zip-packed question banks (txt/md/json/docx), categorizes items across four dimensions (type, difficulty, knowledge point) plus three hierarchical levels (grade, subject, knowledge point), and performs similarity detection. It supports paper generation for daily practice, weekly tests, monthly assessments, midterm/final simulations, and topical training. It also generates personalized compilations based on student weaknesses, mastery levels, and score ranges. Each question is annotated with school, paper type, year, and original question number for full traceability. No API key required.
Read Moredsh-settings-button: Persistent settings entry next to the account in the sidebar
Developed by lindingkao, the client plugin **dsh-settings-button (0.2.0, MIT)** injects a gear and "Settings" button into the right side of the account avatar at the bottom of the DSH Web UI sidebar, matching the height of the user column. A single click prioritizes invoking the host's `settings.open` command to navigate directly to the settings panel, falling back to the account menu if the command is unavailable. The button automatically hides when the sidebar is narrow (56px) and does not duplicate injection if the host already provides a native button. Based on the official README, this article covers installation commands, compatibility, and implementation details, while emphasizing the importance of reviewing the source code and permissions before installation.
Read Moredsh-chat-gutter: Dual-sided Marker Bands and Quick Navigation in the Chat Area
dsh-chat-gutter is a pure client-side plugin for DSH maintained by lin114514-desng (v0.2.29, MIT license). It draws turn number markers, process badges, and response segment arcs on both sides of the chat column, and provides a two-column "Quick Move" panel in the input area that allows navigation categorized by AI questions, tools, AI responses, and user messages. It does not read session data or modify the official DOM structure. The configuration includes a master toggle for the marker band and highlighting of response start points. For installation instructions, see INSTALL-PLAN.md. Requires DSH >= 0.2.1-rc.2.
Read MoreXuanjian Astrology Master Workbench: Integrating 10 Major Divination Systems into a Single Skill
The Xuanjian Mingli Master Workbench (@user_358fb004/xuanjian-mingli-master-workbench) is a community-based "Skill" on SkillHub, classified under "life-service," which integrates ten divination systems: BaZi (Four Pillars), Zi Wei Dou Shu, Qi Men Dun Jia (complete disk transmission), Liu Yao, Mei Hua Yi Shu, Tarot, Western Astrology, Numerology, BaZi compatibility, and BaZi naming. It addresses challenges in recognizing natural language requests, scheduling chart calculations across systems, cross-validation, and generating unified, fused reports. The Skill autonomously handles intent routing, precise solar term and horoscope calculations, four-tier report generation, user profiling, and decision assistance, meaning integrators simply need to pass through user messages. Documentation indicates `requires_api_key=false`, meaning no additional keys are needed. While the official description claims it is a "Production-ready V1.3," the SkillHub page registers the version as 1.0.0; this inconsistency suggests prioritizing the page data for integration. With a page rating of 100000, 956 downloads, and 0 installs, community verification is limited. The documentation lacks installation commands, configuration field details, client-specific instructions (Cursor / Codex CLI / Claude Code), or reproducible command-line examples. Consequently, this text focuses only on capability descriptions, explicitly flags unverified items, and avoids speculation. It is recommended to conduct tests covering single-system charting, multi-system cross-validation, and decision-making requests before integration. Integrators must also take responsibility for compliance boundaries regarding divination content and the precision validation of chart calculation algorithms.
Read MoreViral Video Replication: A Skill for Reproducing Hit Talking Head Videos via Link
"Baokuan Video Replication" is a content creation Skill published on SkillHub, located at @org-qvypbtd6/baokuan, version 1.0.0, published and maintained by the enterprise account "Jingge Technology", categorized as "content-creation", sourced from "enterprise", and requires no API Key. The problem it solves is singular: enter the link of a viral spoken-content video to replicate the corresponding spoken content, eliminating the time required for manual line-by-line deconstruction of the script. The official description reads: "Skills that can rapidly replicate viral spoken-content videos, complete the replication of a viral spoken-content video simply by copying the video link." The page currently shows zero installs, indicating it is recently launched with no available community user feedback. It is recommended to conduct a small-scale trial before integrating it into production workflows. This article does not fabricate local installation commands, configuration fields, or reproducible examples, but only documents its verified capabilities and limitations.
Read MoreSimulating Wang Yuquan with AI: Organizing Wang Yuquan's Public Methodology into a Traceable Knowledge Base
Simulate Wang Yuquan with AI | AI-HIVE is an Agent Skill classified under content creation, released by the AI experts at SharpKey Technology (Version 1.0.0). It first establishes a sample library based on Wang Yuquan’s public sources, then distills his core themes, audience pain points, argumentation structures, content pacing, and visual directions. It can organize articles, speeches, and course notes that users have usage rights for into a methodological knowledge base, providing strategic consultant-style analysis with traceable citations. Through AI-HIVE, it generates original scripts, covers, illustrations, voice-over plans, and short video schedules. It is suitable for entrepreneurs, business executives, professionals, and audiences interested in investment and business content. Usage precautions: Do not clone voices or portraits, do not fabricate experiences or quotes, do not imply authorized partnerships, and do not reproduce copyrighted expressions. Official materials do not provide directly verifiable installation commands; the interface tag is requires_api_key=false.
Read Moredsh-agent-lab: Agent Behavior Regression Lab
dsh-agent-lab (Limtide, v0.4.0) is a DSH workflow plugin that records sessions as runs, enabling side-by-side comparison of tool sequences, parameters, results, and usage. It supports isolated copy replay, task × variant matrices, and CI integration via assertions (e.g., no-tool, file-contains) and agent-lab-check CLI exit codes. Runs and assertions are stored in $DSH_HOME/agent-lab. Install using `plugin_manager install_bundle`. It is ideal for behavioral regression testing and comparative archiving when switching models, prompts, or plugin versions.
Read MoreMicro-Drama Screenplay Creation: Transforming Script Generation into a Standardized, Payable Production Line
An Agent Skill designed for short drama video scripts, located at @user_a75f2ccf/short-drama-script (Version 1.0.0), maintained by a community publisher and categorized under content creation. It is triggered by user requests such as "write a drama script," "short drama script," "plot twist," or "funny short video." The skill requires three inputs: theme, duration, and character settings. Upon payment of 14.9 CNY (via Alipay, no API Key required), it generates a complete script covering storyboards, dialogue, subtitles, camera movements, and soundtrack suggestions, as well as including twist designs, satisfying pacing, and episode outlines. The article outlines its positioning, verified capabilities, prerequisites for activation, typical workflows, and applicable scenarios, while noting that the payment process needs to be integrated into automated pipelines and that compliance risks regarding the community source must be independently reviewed.
Read MoreWhat to Eat Today: Agent-oriented Recipe Generation and Meal Planning Skill
Skill Intro is a life-service skill released by user_3dfd61cb, version 1.0.0, API key required for invocation. It breaks down the "What to Eat Today" problem into three modes: generating simple recipes from ingredients on hand, generating random recipes based on 80% historical preferences and 20% new exploration, and random takeout recommendations. It also logs allergen and dietary restriction. Recipes include images, search. The skill guides the user through setting preferences upon first use. This article is structured according to official documentation, covering its positioning, the three modes, reproducible examples, applicable limitations. It also reminds that installation commands and host environment variations are not provided in the materials and should be verified on the official page before implementation.
Read Moredsh-bsk-daemon: On-Demand Start/Stop of the BrowserSkill bsk Daemon in DSH
dsh-bsk-daemon is a DSH networking tools plugin maintained by like999888999-ui. It solves the issue where BrowserSkill fails to launch bsk on demand (os error 5) on Windows due to the DSH host Job Object. On Windows, a scheduled task and bsk-shim wake the daemon on demand and let it exit after being idle; it provides tools such as bsk_status, bsk_setup, bsk_wake, and bsk_stop. The installation command is `dsh plugin --profile desktop add github:like999888999-ui/dsh-bsk-daemon`, and requires Node ≥ 22.13.0 together with the browser-skill plugin.
Read MoreXiaoguo JoinQuant Quantitative Trading Assistant: Converting JoinQuant Strategy Requirements Directly into Python Code
A Skill for generating and assisting with quantitative strategies for the JoinQuant platform, released by community author Xiaoguo. It takes natural language requirements, generates Python strategy code runnable on JoinQuant, and offers guidance on API usage, helps users troubleshoot errors, and provides suggestions for strategy optimization. It supports triggering via keywords such as 聚宽, JoinQuant, JQ, dual moving averages, grid trading, factor stock screening, Alpha factors, multi-factor, financial data, JoinQuant API, and JoinQuant errors. It requires no API key. Hosted on SkillHub at @user_676c825d/xg-joinquant-trader (version 1.0.1), the article positions it as a tool for writing code and answering questions rather than placing automated orders. It outlines a five-step workflow from requirement description to backtesting troubleshooting, along with best practices regarding its execution environment, description precision, and data coverage.
Read MoreGrant Application Writing Skill: Research Plan, Budget Justification, and Project Argumentation
Grant Proposal Writing (Grant Proposal Writing, @user_3c6cb52e/grant-proposal, v2.0.1, community-sourced, professional category) is targeted at the writing of grant application materials. It integrates three parts—research plan, budget statement, and project justification—into a unified writing framework to avoid inconsistencies between timelines and funding calibers within the application materials. The tag indicates requires_api_key=false, so no additional API Key configuration is required. This article clarifies its positioning, the verified core functions, the official address, and the precautions for activation, and points out truthfully that installation commands and examples are missing; therefore, the official page should be treated as the source of truth.
Read Moredsh-direct-open: Enable opening files in DSH replies with the default system application via left click
dsh-direct-open is a pure client-side DSH plugin maintained by likaileave that intercepts clicks on delivery file cards, modified card file lines, and body text file references. Left click opens files with the system default program or locates them in the file manager, modifier keys like Shift allow the original sidebar behavior to pass through, and right click provides a menu with options to open, locate, or copy path. Settings are located under "Settings → Open With" and stored in localStorage. Requires DSH 0.2.0-rc.2+ and desktop openWorkspacePath capability. Installation example: dsh plugin --profile desktop add dsh-direct-open.
Read MoreBudget Execution Management Expert: Incorporating Budget-Actual Variance into Closed-Loop Control
The Budget Execution Management Expert is a community Skill (identifier: @user_ce145526/ba-capability-101, version 1.0.0, category: business-ops). It dynamically tracks the budget-to-actual performance for revenue, costs, and assets. The tool monitors variances in real time, analyzes deviation causes, and implements timely corrective adjustments, thereby strengthening closed-loop control. This capability optimizes resource allocation and maximizes asset efficiency, and it requires no additional API Key. The article details its positioning, verified capability scope, installation and activation procedures, and available reference information. It also presents a general usage workflow based on the introduction. Additionally, the article outlines certain limitations, noting that official documentation lacks reproducible examples, and that API fields and threshold rules require independent verification.
Read MoreData Visualization Report Generator: From CSV/Excel to Interactive HTML Reports
This article introduces the "Data Visualization Report Generator" Skill (Coordinate: @user_5735bd90/data-viz-report, Version 1.0.1, Category: data-analysis) released by the SkillHub community. It generates interactive HTML reports from CSV/Excel data, supporting multiple templates, Chinese-English language switching, one-click screenshots, and filter linkage, without requiring an API Key. The article outlines core features, trigger methods, typical usage, and applicable scenarios. It also notes that the official page scraping currently failed and the installation command was not provided in the materials, suggesting users verify the details at the official address before use.
Read Moredsh-board: Managing Session Tasks using Kanban in DeepSeek Harness
dsh-board (@lihuu/dsh-board) provides per-workspace Kanban for DeepSeek Harness: with configurable columns for To Do/In Progress/Done, task cards bound to persistent Sessions, badge indicators showing real-time status such as Running, and actual conversation.content embedded within the board. It supports Start-triggered automatic transitions with time tracking, sidebar overlays with keyboard navigation, and state persisted to ~/.dsh/dsh-board/state.json. Installation: dsh plugin add @lihuu/dsh-board and restart; requires DSH ≥ 0.2.2. MIT licensed, maintained by lihuu.
Read MoreHR: AI Skills to Free Staff from Repetitive Tasks
"Renshi (HR)" is an enterprise HR-oriented AI Skill published by Kunlun Growth on SkillHub (located at @kunlungrowth/kunlun-renshi, version 1.1.0, category business-ops, source enterprise, requires_api_key=false). It consolidates four types of highly repetitive HR tasks—recruitment JD generation, resume initial screening, interview outline preparation, and employee onboarding & policy Q&A—into callable workflows, enabling the model to produce first drafts that are subsequently reviewed by humans, thereby reducing drafting costs. This article, compiled based on official scraped materials, covers: the Skill's positioning and publication details; a point-by-point explanation of its four core capabilities; prerequisites for activation (search and install by coordinate, no additional API Key required; since no specific commands were provided by the official source, differences among Cursor/Codex CLI/Claude Code are not speculated); typical usage paths inferred from functional positioning; applicable scenarios and caveats, including the current installation count of 0, the need for manual review of resume screening, and the requirement to connect internal enterprise policy texts for policy Q&A. The official URL is appended at the end of the article.
Read MoreParent-Teacher Conference Communication Assistant: Turning Meetings into Well-Prepared Discussions
The "Parent-Teacher Conference Communication Workflow Assistant" is a community educational Skill on SkillHub, published by user_fa14b5bc. It is at Version 1.0.1, sourced from the "community" channel, categorized under "education", and holds a rating of 100,000 with 963 downloads and 0 installs. The tag `requires_api_key=false` indicates no API Key is required. By inputting the child's grade and the parent's focus areas, the tool automatically generates a comprehensive parent-teacher conference workflow plan. This includes a pre-meeting checklist of 5 to 8 specific questions, response scripts for four types of teacher feedback scenarios during the meeting, post-meeting parent-child conversation scripts tailored to elementary, middle, and high school levels, as well as a semester follow-up plan. The article details its core features, installation and activation methods (via `skillhub install` coordinates or manual directory specification), typical usage, applicable scenarios, and precautions.
Read MoreDSH Plugin Steward: Integrated Management for Update Checks, Automated Upgrades, and Rollbacks
dsh-plugin-steward (Plugin Steward) is a DSH client plugin maintained by lihao178, providing a left-panel UI and a dsh-pm CLI. With a zero-dependency Node.js implementation, it handles plugin update checks, pre-install backups with rollback on failure, static security warnings, health checks and repairs, duplicate detection and cleanup, and registration of Chinese display names and usage notes. It supports dual update sources—npm and GitHub Releases—making it suitable for local Harness environments with many plugins that require automated operations. Installation: `dsh plugin --profile desktop add github:lihao178/dsh-plugin-steward`.
Read MoreVisual PRD: Unified Offline Review of PRDs, Prototypes, and Business Annotations on a Single Screen
Visual PRD is a community-sourced Agent Skill with the identifier @indiv-ai_fanche/visual-prd-aifanche, version 4.0.2, categorized under office-efficiency. It is published by user_586f97f0 and tagged with requires_api_key=false. This skill assembles existing or newly generated PRDs, interactive prototypes, and business annotations into a single-screen offline review page: it can generate a review starting point from requirements, and can also build verifiable annotation data and page anchors from existing HTML before assembly. It is suitable for requirement reviews, prototype annotation, and enhancing existing HTML. It is not intended for simply writing ordinary PRDs, nor does it deliver final high-fidelity visual deliverables. The official page does not provide a copyable installation command; it must be installed via the client Skill marketplace by searching by identifier or name.
Read More