AI Project Delivery: A Cross-Harness Methodology
An Agent Skill and DSH plugin maintained by anjiecobamboo-blip. It provides an eight-stage workflow from requirements alignment to delivery retrospective, complete with diagnostic tables, decision heuristics, and school-of-thought divergences, plus a non-technical user mode. Supports installation via `npx skills`, loading from the DSH skill catalog, or profile plugin package loading. It is a non-executing pipeline that can be combined with Spec Kit, BMAD, and others. MIT-licensed, v0.1.5.
Read MoreLong-Form WeChat Article Creation: Breaking Down the Writing Process into Six Stages
WeChat Official Account article creation (@clawhub_fengbabao0929/wechat-article-writer, v1.0.0, content-creation, MIT-0) is a writing assistant for WeChat Official Accounts on ClawHub, progressing through six stages: topic selection → title → framework → draft completion → save → image prompt generation. Topic selection defines the theme and target audience; title optimization must provide 5-10 options and wait for user selection; after framework construction, write a full 2500-3000 word article; format in Markdown with automatic saving; then generate cover, infographic, and guidance image prompts, only calling the script to generate images after confirmation. Writing style is constrained by writing_style.md: conversational, complete sentences, pragmatic. Installation uses `openclaw skills install @fengbabao0929/wechat-article-writer`, with trigger words: 写公众号, 公众号文章, 推文, /wechat. Note that the save path is hardcoded as a Windows path, and image generation depends on the API Key in scripts/image.env.
Read MoreIT Opportunity Radar: Early Detection of Xinchuang Projects - Capturing Leads Prior to Project Initiation
This is an Agent Skill designed for discovering early-stage business opportunities in the IT industry, maintained by community publisher user_a15e4142. It is categorized under "business-ops" (Version 1.0.8) and requires an API Key for activation. It provides four types of leads: queries for planned digital projects (Smart City/Digital Government project initiation phases, 6-18 months in advance), IT procurement intentions (1-3 months pre-tender), expiring maintenance and software service contracts, and client follow-up suggestions ranked by budget and urgency. By inputting a technical domain or region, it outputs a value-ranked list of business opportunities. The trigger conditions do not require users to mention IT explicitly, covering sectors such as IT Application Innovation (Xinchuang), Localization, Software Development, System Integration, Cloud, Data Centers, Cybersecurity, and Smart City. It is suitable for IT sales, bidding support, and market research personnel. Limitations include the need to independently verify the advance notice period and the non-disclosure of ranking weight algorithms.
Read MoreResume Master: Diagnosis, Matching, and Rewriting of Chinese Job Resumes
For a long time, resume optimization for Chinese-speaking job seekers has relied on two types of tools: general large language model chatbots, which provide fragmented answers without a fixed workflow; and "one-click beautification" features on recruitment platforms, which only adjust layouts but fail to improve the actual description of experiences. The real bottleneck lies in a single issue: how to align one's experience with job requirements, and how to phrase that experience into compelling sentences. "Resume Master" is an Agent Skill designed for Chinese-speaking job seekers, integrating resume diagnosis, JD matching, experience rewriting, career transition adaptation, ATS checks, and interview story preparation into a reusable workflow. The following sections introduce its capabilities, activation methods, and practical usage. **What is this?** Resume Master is an Agent Skill published and maintained by community developer user_cae83fba, located at @user
Read MoreChobits-Chii-DSH: Applying the Chii Picture Book Style Skin to the DeepSeek Harness Web UI
chobits-chii-dsh is a purely client-side skin plugin by Anime2Real for DeepSeek Harness Web UI, featuring neutral, warm-toned light/dark tokens, a Polaroid-style photo wall, an optional Live2D Chii desktop pet, and hot-unload. Based on the official README, this article introduces its features, npm/GitHub installation commands, Discover index sources, desktop pet configuration, and engines.dsh compatibility, and explains that the code and assets are sublicensed under the MIT license and that reviewing the source code before installation is necessary.
Read MoreVideo Visual Style Library: Converting 16 Product Infographic Styles into HTML/CSS Video Templates
Sourced from the SkillHub community, the Video Visual Styles Library (@user_d8831e99/video-visual-styles, v1.0.0, category: design-media) converts the 16 product infographic styles defined by product-infographic-styles (e.g., handmade paper craft, 3D clay, blackboard chalk, Memphis, lab blueprint) into HyperFrames-renderable HTML/CSS templates. It integrates these as a video visual style dimension within a modular video engine. Trigger phrases include "video style", "visual style", "change video style", "product infographic style", and "16 styles". No API key is required. The article outlines its capability boundaries, integration prerequisites, invocation approaches, and precautions.
Read MoreFlight Master - Flight Status Service: Delegating Flight Status Queries to Agents
The Flight Manager - Flight Status Service is a life-service Skill (v1.0.3, life-service) published by community developer user_594f630b. It exposes flight status tools via MCP Streamable HTTP, covering airport/city search, flight inquiry (preceding flight, schedule adjustments, crowding level, on-time performance, fare calendar), aircraft type and cabin class, weather, itinerary check-in, and value-added services such as ticket refunds, changes, and rebooking. The Skill auto-activates when users discuss flights, airports, weather, or check-in and ticket change/refund matters. The article first outlines the pain points of self-built flight data pipelines, then explains the Skill's positioning and capability list. It subsequently specifies the activation requirements (an API Key is required; access is provided through MCP Streamable HTTP, with the endpoint and configuration fields subject to confirmation on the official page) and demonstrates typical usage by delegating queries to the tool. Finally, it lists applicable scenarios and points to note.
Read Moredsh-window-drag: Restore title bar dragging and double-click zoom in DSH Desktop (macOS)
dsh-window-drag is a DSH client plugin maintained by angls4, which only works on DeepSeek Harness Desktop for macOS. When some builds ignore the webkit drag area, causing the window to be undraggable, it enables dragging on the top bar (48px) and sidebar header via APIs such as moveTo/resizeTo. It also supports double-clicking to zoom in and restore within the workspace, and includes a startup rescue function for windows that have moved off-screen. It can be installed via the plugin manager bundle github:angls4/dsh-window-drag without requiring a restart. Please review the source code and the MIT License before installation.
Read MoreExcel Report Automation Expert: Transforming Data Reporting from Manual Compilation to Automated Generation
Addressing the pain points developers face during data reporting, this article introduces the Excel Report Automation Expert Skill: intelligent report generation (one-click data analysis and automatic chart selection), data insight analysis (automatically identifying trends, outliers, and key findings), template application (preset templates for weekly reports, monthly reports, sales dashboards, etc.), custom chart generation, multi-table summary dashboards, and automatic formatting beautification. Published by community developer user_92725383, sourced from community, categorized under office-efficiency, with coordinates @user_92725383/excel-report-automator, current version 1.0.0, rating 100000, downloads 1080, and tags requires_api_key=false (no API key required). It is suitable for scenarios where quick report, dashboard, or analysis report generation is needed after obtaining data. The article details functional boundaries, activation methods, and items to be confirmed (the official page does not provide installation commands and dependencies), and provides the official address.
Read Moredsh-codex-effort-switcher: Codex-style Model and Reasoning Effort Selector for DSH Composer
dsh-codex-effort-switcher is maintained by AndyZh0 and released under the MIT license (currently 0.2.2). It mounts a Codex-style card — with an effort tab, a segmented slider, and a model list grouped by Provider — into the DSH Composer model slot at priority -1. It still reads and writes the same modelDirectories, and the behaviors of /model, the Models setting, and session.selectModel remain unchanged. To install, run npm build and pack in the source directory, then execute dsh plugin add tgz and restart the application. It also provides codexModelTiers for third parties to register custom effort tiers.
Read MoreDe-AI-ify Skill: Detect AI Tone, Pinpoint the Issues, Rewrite to Sound Human
De-AI-ify is a Chinese-text “de-AI-ification” rewriting skill released by RealQ and labeled with the source as enterprise. It breaks de-AI-ification into three steps: detecting an “AI flavor,” pinpointing specific issues, and rewriting into more natural expressions while preserving facts and logic. The skill follows the standard SKILL.md format, requires no API key, and is currently at version 1.0.0 under the content-creation category. It is suitable for polishing technical documentation, adjusting product copy, and correcting tone in everyday communication, and can also serve as a post-processing step in a writing workflow. Rewriting is limited to language style and does not involve fact checking.
Read Moredsh-claude-ultra: Enabling Claude-style Ultra Reasoning and Multi-Agent Orchestration in DSH
`dsh-claude-ultra` is a DSH workflow plugin maintained by AndyZh0 that adds an "Ultra" fifth tier to the dsh-codex-effort-switcher slider. It maps the model's reasoning effort to the maximum level and injects the Claude ultracode system prompt. Requiring switcher 0.2.0+, it is installed via local `npm pack` and `dsh plugin add`. Supporting `ultracode` and `ultrathink` keywords along with session persistence, it is designed for developers who need to align Claude Code's maximum-level reasoning with subagent orchestration.
Read MoreJiuQuanEr Unified Fund Data Query: Integrating 7 categories of fund data into a single Skill
"Jiuquaner Unified Fund Data Query" is a Skill from the community source (@user_24b33f9d/jiuquan-skills, version 1.2.3, categorized as data-analysis). It integrates seven fund data modules into a single entry point: Fund Manager Profiles (person-desc11), Historical Fund NAV (jz), Fund Details (get-fund-info), QDII Subscription/Redemption Limits (fund-sx-qdii), Jiuquan Fear & Greed Index (kt), Equity-Bond Valuation (gz/bond/fed/temp, including China 10-year government bond yield, market temperature, FED Risk Premium/Dividend Yield Risk Premium), and Index Valuation Lists (indexvaluation/industry/group-list, providing PE, PB, dividend yield, ROE valuation percentiles, and interval returns categorized by industry, theme, strategy, scale, selection, overseas, and others). It is activated by intent keywords, allowing users to trigger the corresponding modules via natural language questions without needing to write manual invocation statements. The resource is tagged with requires_api_key=false. The official page failed to be scraped this time, and the installation command could not be verified; the resource shows a rating of 100,000, 1,180 downloads, and 0 installations, which is inconsistent with the 99,208 and 1,002 indicated in the clues. It is recommended to re-verify before citing.
Read MoreInstructional Design for Integrating Ideological and Political Education: The "Autumn Leaf Workplace" Case Study
This document introduces the positioning, trigger words, source information, and integration methods for the "Qiuye Workplace Skill - Introduction to Curriculum Ideological and Political Education Teaching Design." Released by Qiuye Group and categorized under education (Version 1.0.0), this skill requires no API Key. It is designed to generate teaching design plans incorporating ideological and political elements, with the trigger words being "Curriculum Ideological and Political Education Teaching Design" and "Teaching Design Integrating Ideological and Political Elements."
Read More@andrewhmir/dsh-plan-review: Read the plan in Composer before clicking approve.
The community plugin `@andrewhmir/dsh-plan-review` is designed for the DeepSeek Harness Web GUI. It takes over the Composer during superpowers' `[plan-review]` prompts to render full plan Markdown inline, featuring a top bar with Accept/Revise buttons, text selection commenting, and session draft support. It requires DSH >= 0.2.1-rc.2 and the `superpowers-ui` skill. Install via `plugin_manager install_bundle` and restart `dsh web`. Plan files must be located within the workspace root and are read-only by the host; when comments are present, "Revise" passes feedback back via `custom`. MIT licensed; unofficial bundle.
Read More**Stock Market Narrative Veracity Detection: A Seven-Stage Audit of Bullish Narratives**
The article introduces the Skill “Stock Market Narrative Truth Detector” (v2.0.0, professional, requires_api_key=false), published by community developer user_5a465dbe. It decomposes the question “why did this stock rise” into a seven-stage audit framework: Narrative System Construction → Fundamental Anchoring → Matching Score → Evidence Grading → Capital Alignment → Logical Reasoning → Full-Lifecycle Early Warning. The goal is to distinguish whether a rally narrative is built on solid fundamentals (rock) or hollow concepts (sand). The article explains its positioning, trigger words, runtime conditions, and how to organize tasks by trigger words. It also outlines limitations such as reliance on evidence quality, the method’s sources being unverifiable, missing installation instructions, and zero installs.
Read MoreDashboards and Design Guide for BI Tool Selection: Practical Implementation Approaches from Tool Selection to Metrics Alignment
The *Data Dashboard and BI Tool Selection Design Guide* is a methodology-driven Skill produced by Kunlun Growth (Version 1.0.0). Classified under data-analysis and sourced from enterprise, it covers the BI tool selection decision framework (self-built open-source, commercial SaaS, reporting plugins), dashboard information architecture based on "one screen, one decision," chart type selection references, metric definition governance, and permission design. It includes dashboard requirement interview templates and go-live acceptance checklists to address common issues such as unused dashboards, metric inconsistencies, and incomprehensible charts. It does not provide data connection or rendering capabilities and must be implemented in conjunction with BI tools. It is suitable for data analysts, BI engineers, and managers driving data culture construction. Official documentation does not provide installation commands or explanations regarding differences between coding agents. Tag: requires_api_key=false.
Read Moredsh-qqbot-refine: Frequency Gates, Personas, and Local Governance on QQ Channel
An introduction to the DSH plugin dsh-qqbot-refine maintained by Anarrogantcat: it implements frequency and privacy gates ahead of the model (commands and pushes incurring 0 tokens), supports onebot/tencent dual channels, and integrates persona, half-life memory, local mood state, relationship intimacy, group administration, and capabilities such as image generation/TTS/knowledge base. It also covers NapCat configuration, the install command (npm install -g dsh-qqbot-refine), and the panel path (Main Console → Plugins → dsh-qqbot-refine), with reminders on process privileges and local storage boundaries.
Read MoreHarmonyOS NEXT Pro: A Skill Covering the Entire Development and Deployment Workflow for HarmonyOS NEXT / OpenHarmony Applications
HarmonyOS NEXT Pro is a community-maintained Skill based on the universal SKILL.md format (Coordinates: @user_8fb2e41b/harmony-next-pro, Version 1.2.0, Category: dev-programming, No API Key required). It targets the entire lifecycle of HarmonyOS NEXT / OpenHarmony native application development, covering review, debugging, migration, building, deployment, and automation. The Skill covers ArkTS strict mode; ArkUI declarative components and animations; the Stage Model (UIAbility/ExtensionAbility/AbilityStage/WindowStage); configurations for module.json5, app.json5, oh-package.json5, and build-profile.json5; two sets of state management decorators—V1 (@State/@Prop/@Link/@Provide-@Consume/@Observed+@ObjectLink/@Watch) and V2 (@ComponentV2/@Local/@Param+@Once/@Param+@Event/@ObservedV2+@Trace/@Monitor)—as well as AppStorageV2/PersistenceV2. It also addresses Navigation+NavPathStack routing, LazyForEach/@Reusable performance optimization, 60+ HarmonyOS Kits, TaskPool/Worker and @Sendable, NDK/NAPI, ArkWeb CDP, and DevEco Studio tools (CodeGenie, MCP, ArkUI Inspector, Profiler, Previewer). Furthermore, it includes diagnostics using hdc/uitest/aa/bm/hilog/hidumper and the harmony-deploy CLI for HAP/HSP/HAR packaging, signing, multi-module dependencies, product variants, debug/release/test builds, and AppGallery App Packaging. Additionally, it provides offline API reference routing via KITS/TASK_MAP/INDEX. This Skill is designed for users handling HarmonyOS projects within generic Agents that support the SKILL.md format. For ratings and download counts, please refer to the real-time data on the SkillHub page.
Read MoreEquity Incentive Query: Connecting Listed Company Equity Incentive Plans and Share Repurchase Data to Cue API
Equity Incentive Query (@org-y1eh5fyg/cue-equity-incentive, v1.5.10) is maintained by Beijing Ganyee Intelligence Technology Co., Ltd., categorized as 'professional,' and tagged with 'requires_api_key=true'. Its official repository is located at https://api.skillhub.cn/org-y1eh5fyg/cue-equity-incentive. It connects queries for listed company equity incentive plans and share repurchases to the Cue Data Interface. Quick queries return results in seconds for individual plans or specific elements, or for comparable elements across multiple companies simultaneously. For plan design, implementation effects, and peer comparisons for a single company, as well as market-wide plan inventories and trend analysis, it generates in-depth reports powered by free-style deep research. The data foundation officially claims to cover over 10 years of history back to 2015, serving as an exclusive and proprietary data source. This article outlines the problems it solves, verified capability boundaries, key installation and activation points (requiring a Cue API Key, with other configuration fields based on the official page), and typical usage, while also explaining applicable scenarios and potential pitfalls.
Read Moreprompt-free-047: Batch Generation Prompt Templates for In-Feed Ad Copywriting
In information feed advertising placement, three issues are commonly encountered: creatives are consumed quickly, copy becomes homogeneous, and delivery volume fluctuates. Writing each piece of copy manually is costly, while using off-the-shelf large models via direct prompting often fails to ensure consistent style or adherence to platform norms. prompt-free-047 is a Skill from the SkillHub community, located at @user_2c08c7ce/prompt-free-047 (version 1.0.51, category: content-creation). It is free and does not require an API key. By turning the batch generation of information feed ad copy into built-in prompt templates, it lets users provide product information, selling points, target audience, placement platforms, and the desired quantity, and then receive multiple copy variants in a unified structure without needing to rewrite prompts each time. This article explains its core features, how to enable it, and usage examples, and notes that the generated results still require manual review in accordance with each platform’s rules and advertising compliance requirements; it is not intended to replace human judgment on selling-point direction and audience targeting.
Read Morehtml-visualizer: Turns requirements directly into openable single-file HTML artifacts
html-visualizer is an Agent Skill that transforms content requirements into single-file HTML artifacts, published by user_5b589a79 in the SkillHub community under the dev-programming category, version 1.0.0. It automatically routes requirements to types such as interactive courseware, data visualizations, UI prototypes, flow diagrams, and AI tool guides, producing standalone HTML pages that are immediately useful on the first screen, responsive, and dark-mode-friendly. It includes 20+ built-in quality validation scripts and STEM lecture courseware templates. This article explains what it is, its core capabilities, how to enable it, the typical workflow, applicable scenarios, and limitations.
Read Moredsh-disk-manager: Browse Windows disks and workspace files in the DSH right sidebar
dsh-disk-manager is a DSH client plugin (v0.4.2, MIT) maintained by Ananpng123, designed for Windows 10/11. It provides volume overview, directory browsing, address bar navigation, on-demand folder size calculation, hover preview, desktop inline layout, and a workspace document list grouped by session within the right sidebar. Write operations are limited to deleting to the Recycle Bin and creating new folders; partition formatting is not supported. Installation requires adding the index `https://ananpng123.github.io/dsh-disk-manager/index.json` to the Plugin Market or using `plugin_manager install_bundle`, followed by a DSH restart to enable `/api/dsh-disk`. Requirements include DSH >= 0.2.1-rc.5 and PowerShell 5.1.
Read MoreCinematic Camera Movements Script Generator (Dòu Zhìgāng (Victor Dou) Style Movements): Breaking down short video camera movements into executable scripts
A Community Skill based on the universal SKILL.md format, the Cinematic Camera Script Generator (Dou Zhicang-style camera movement) is identified by the coordinates @user_ebee0fcc/cinematic-camera-script, version 1.0.1. Released by user_ebee0fcc and categorized under content-creation, it transforms short-video camera movements from descriptive adjectives into executable scripts. By inputting images or text descriptions, it outputs a complete script containing second-by-second storyboard breakdowns, camera movement parameters, and BGM beat-syncing rhythms, along with one-click video prompts. The results can be used as reference for live-action shooting or for AI video generation. As the source material does not provide specific installation commands, directory paths, or configuration fields, nor verify the interface differences among Cursor, Codex CLI, and Claude Code, this section only outlines the input-output format without supplementary command instructions. For usage: prepare reference images or text descriptions to feed into the Skill; execute the second-by-second storyboard breakdown; when shooting live-action, adjust the camera according to the movement parameters; when generating videos with AI, apply the provided prompts; and finally, align the visuals with the music based on the BGM beat points. It is important to note that this tool only generates scripts and prompts; it does not replace actual shooting and editing. Camera movement parameters and beat-syncing results must be verified against the actual equipment, lens, and music. The page shows a score of 100000, 1042 downloads, and 0 installations; these statistics are subject to change over time, with real-time data on the official page being the authoritative source. Official Address: https://api.skillhub.cn/user_ebee0fcc/cinematic-camera-script SkillHub Page: https://www.skillhub.cn/skills/user_ebee0fcc/cinematic-camera-script
Read MoreSemiconductor Financial Analysis Engine: Deconstructing Chip Company Financial Reports via a 6-Step SOP
The Semiconductor Financial Analysis Engine is a professional vertical Agent Skill released by the community, located at @user_238915fc/semiconductor-financial-analysis. Currently at version 1.3.8, it requires no API Key. Designed for financial analysis within the semiconductor industry, it covers six major tracks: IC design, wafer fabrication, semiconductor equipment, semiconductor materials, memory, and packaging & testing (OSAT). Its core design follows a fixed six-step SOP: external climate prediction → competitiveness mapping → four-dimension financial breakdown → quantitative risk identification → climate prediction → conclusion. This structure places industry sentiment assessment at the beginning and end, while performing financial breakdowns and risk identification in the middle, ultimately converging on a conclusion. The input consists of financial reports and industry research reports provided by the user, rather than generated from scratch. The output boundaries are clearly defined: it performs financial analysis exclusively and prohibits valuation, stock prices, price movements, bull/bear outlooks, or investment advice. It serves as a mandatory linkage engine for semiconductor-equity-investment-dd. It is recommended to install it alongside semiconductor-equity-investment-dd and equity-valuation to form a closed loop for primary market investment analysis; however, it can also be installed independently for financial due diligence and financial report interpretation of semiconductor targets. The official side does not provide a one-line installation command or client-specific instructions; installation can be performed on the SkillHub page using the coordinates or address. Users must prepare the original financial reports and industry research reports themselves and are advised to test-run and verify the skill on a real financial report before applying it to formal workflows.
Read More