Image Search Assistant: Find real-life photos for Chinese content.
The Search Assistant is an Agent Skill based on the general-purpose SKILL.md format, maintained by publisher user_8170ea13 of the SkillHub community (Version 1.1.4, Classification: design-media). Targeting scenarios such as WeChat Official Account covers, website banners, and Xiaohongshu illustrations, it retrieves real photographs rather than AI-generated images. Its core positioning prioritizes real-world scenes in China, empowers users to select images, performs visual verification, and provides hierarchical licensing annotations. It aggregates domestic sources including Baidu Images, 360 Images, Duitang, and Sogou Images, requiring no API Key. It features overseas backup sources, Pexels and Pixabay, for which users must configure their own Keys. The skill supports filtering by size (large, medium, small) and orientation (landscape, portrait), as well as style guidance keywords via the `--style` parameter. It ultimately generates a gallery for user selection and annotates the licensing level. When user queries are ambiguous but imply a need for Chinese real-scene photos, illustration retrieval, or copyrighted images, the Skill proactively triggers guidance without requiring precise command keywords. It is not applicable for image generation, AI drawing, or reverse image search. Whether an image can be used commercially is subject to the annotation level and final confirmation by the user. No executable installation commands are provided in the documentation; users should refer to the official page for instructions.
Read MoreHistorical Figure Story Creation Tool: Turning History Books into Readable Stories
A general-purpose Skill for story-based scenes centered on historical figures, published by SkillHub community publisher user_32475b80 at @user_32475b80/history-figure-story. Version 1.0.0, categorized under content-creation, sourced from the community. Core capabilities include: rewriting biographies into easy-to-understand and engaging plain-text story scripts; producing presentations at different difficulty levels tailored to three reader groups—general audiences, adolescents, and children; writing short history-popularization articles; and clearly distinguishing between historical facts and fictional content in the finished pieces. Officially listed trigger scenarios include “Tell me the story of XXX,” “Historical figures written for kids,” “Short history explainers,” “Biographies/paragraphs of notable figures,” “Teach Su Shi to middle school students,” and “Make history fun,” among others. It is marked as requires_api_key=false, meaning it can be enabled without configuring any API keys. The output is purely text story scripts, without images, audio, or layout/typesetting. It is not responsible for historical source verification, and the quality of the final script depends on the input materials. This article introduces its features, how to enable it, typical prompt examples, as well as applicable scenarios and limitations.
Read MoreJD Direct Curated: Aggregating JD's Six Major Channels into a Single Skill
# Introduction When developing e-commerce shopping guide Agents, two common issues often arise: first, scattered data sources, as JD.com's subsidies, flash sales, and best-sellers are distributed across multiple channels; second, cluttered results, as search results are mixed with third-party stores, compromising product quality and assurance. The "JD Self-Operated Selection" Skill introduced below is designed to address these two pain points. # What is this "JD Self-Operated Selection" is a community Skill designed for life-service scenarios, located at `@user_b95ee7e5/jd-selection-new`. Published by user_b95ee7e5, the current version is 1.2.1. It aggregates six major JD.com channels (Super Subsidies, New Product Launches, Lowest Historical Prices, Flash Sales, 9.9 Free Shipping, and Real-time Best-sellers) into a single entry point, exclusively...
Read MoreZi Wei Dou Shu · Twelve Palaces Charting: A Skill for Charting and Explaining Terminology Based on the Rules of *Zi Wei Dou Shu Quan Shu*
"Zi Wei Dou Shu · Twelve Palaces Chart Casting" is a Skill published by community publisher user_2a1415c1 on SkillHub. It falls under the life-service category, version 1.0.7, with requires_api_key=false. It is completely free, requiring no software installation, no downloads, no command execution, no internet connection, and no account. Following the commonly accepted rules of the "Zi Wei Dou Shu Complete Book," it sequentially completes the following steps: determining the Life and Body palaces starting from the Yin palace by month, assigning palace heavenly stems via the Five Tiger Escape method, determining the Five Elements bureau via Na Yin (tone-pitch), placing Zi Wei based on the bureau number and day, placing the fourteen major stars through the Zi Wei and Tian Fu systems, the Ten Heavenly Stems Four Transformations, the Three-Sides Four-Orthogonal configuration, and the forward/reverse Da Xian (Major Period). It outputs the twelve palaces, fourteen major stars, the palace positions of the Four Transformations, and the Three-Sides Four-Orthogonal relationships, while providing plain-language explanations for each term and its source. Trigger words cover: Zi Wei, Dou Shu, life chart, twelve palaces, major stars, Four Transformations, Flying Stars, Da Xian, annual cycle, Three-Sides Four-Orthogonal, Life Palace, Body Palace, Career Palace, Wealth Palace, Marriage Palace, Travel Palace, occult arts, traditional Chinese culture. It also matches when users ask about annual cycles, job changes, or which year to pay attention to, but it explicitly does not make predictions or judgments. No official installation command is provided in the documentation; under the general SKILL.md format, you can retrieve the SKILL.md from the SkillHub page and place it in the Agent's skills directory to enable it. The Skill explicitly does not judge the quality of palaces, does not make fortune or misfortune assessments or timing judgments, does not provide luck-changing or resolution advice, and does not collect personal information. Different schools may have different charting methods for the same birth chart; it is for reference only and does not constitute a basis for decision-making.
Read Moredsh-plugin-overview: Overview and Organization of Locally Installed Plugins
dsh-plugin-overview is a DSH client plugin maintained by LTao96 that does not handle installation or uninstallation and is used alongside dshmarket. It lists dependencies installed for the current profile under Settings → Plugin Overview, supporting remarks and grouping (state is stored in $DSH_HOME/plugin-overview/state.json), optional AI summaries, read-only health diagnostics, bundle start/stop, and navigation to settings or the main panel. It also allows collapsing third-party settings.sections. **Installation:** `dsh plugin --profile <profile> add dsh-plugin-overview`. Please restart and open it in Settings. Licensed under MIT, Version 0.2.1.
Read Moredsh-kimi-control：Applying Kimi Workspace Interaction to the DSH Sidebar Browser
dsh-kimi-control is a DSH client plugin maintained by lswang9. While retaining all browser tools from dsh-builtin-browser, it injects Kimi Code desktop-style tab strips and toolbars into the sidebar, and displays Agent control and user takeover via status chips. Within the conversation stream, Kimi tool cards are used to present browser steps. Desktop deployment additionally requires running the desktop-bridge installer; requirements are DSH >=0.2.0-rc.2 <0.3.0 and Node >=22.19. Released under the MIT license; this is not an official Kimi/DeepSeek plugin.
Read MoreIntelligent Comparison of Tender Document Versions (Employer Edition): Pre-Addendum Version Reconciliation and Challenge Risk Self-Inspection
This skill is a bidding document version comparator designed specifically for the Tenderee (the party issuing the tender). By uploading two versions in DOCX or PDF format, it compares all changes item-by-item and evaluates challenge risks, compliance thresholds (e.g., deposit ratios, bid submission deadlines, qualification requirements, and relevant legal clauses), as well as terminology consistency from the Tenderee's perspective. It generates a pre-publication self-check report that includes a P0-to-P4 prioritized action list, consistency scans, and item-by-item recommendations. It is ideal for pre-publication self-checks before issuing addenda or correction notices, change tracking during multi-version iteration, pre-emptive risk assessment and response preparation for potential challenges, and procurement archiving. No additional API Key is required. Simply trigger the Agent with commands like "Compare these two tender documents (Tenderee version)" or "Run a self-check before the addendum" to activate. This report serves as a self-check material and does not substitute the final compliance judgment of the Tenderee or their legal counsel.
Read MoreDatabase Design and Optimization: Entrust ER Modeling, Normalization, Indexing, and Sharding to the Skill Specification
Introducing the SkillHub community skill “Database Design and Optimization” @user_3c6cb52e/self-ent-tech-database-design, version 2.0.1, category dev-programming, sourced from the community, with the tag requires_api_key=false. It is positioned as a database design and optimization guide covering ER modeling, normalization design, indexing strategies, query optimization, and sharding and partitioning schemes, aimed at scenarios such as creating tables, schema review, troubleshooting slow queries, and evaluating data sharding. The article explains the problems that each of the five topics aims to solve, and notes that since the official materials do not provide reproducible examples or installation commands, no unverified commands were included; it also points out that there are discrepancies in the metrics for ratings and download counts in the public data, and the figures are based on the records retrieved on 2026-11-05.
Read MoreTeacher's Assistant for Grading and Feedback on Student Graduation Written Thesis: A Guide for Teachers on Initial Review
A Skill for Generating Preliminary Review Opinions on Graduation Theses for University Teachers. Published by community developer user_72b898e0 (Version 2.1.13), categorized under Education, and requires no API Key. Triggered by keywords such as graduation thesis guidance, graduation project guidance, student paper supervision, thesis review, and AIGC rate detection, it generates preliminary opinions on student papers for teacher reference. This article outlines its positioning, core features, installation access, and applicable boundaries, emphasizing that the output is solely for preliminary review reference and does not replace teacher judgment or formal institutional checks.
Read Moredsh-session-suspend: Collapse idle sessions into the sidebar "Suspended"
dsh-session-suspend is a DSH client plugin maintained by Lqg97. It moves idle sessions that are temporarily needed no more into the collapsible “Suspended” group at the bottom of the sidebar, supporting individual or bulk restore as well as suspend-undo. Suspending reuses the host’s archive state and is only effective for idle sessions; records are stored in localStorage. Installation: clone the repository and select the local directory in the plugin manager; it requires DSH 0.2.1-rc.5 and a matching slot. Licensed under MIT, version 0.1.0.
Read Moredsh-egress-audit: DeepSeek Harness Read-Only Data Egress Audit Panel
dsh-egress-audit is a community plugin for DeepSeek Harness maintained by Lqba-dx. It provides a read-only panel under Settings → Data Egress, allowing observation of the host's fetch, Socket connections, and outbound network destinations related to model streams and tool calls. It provides trust grading and suspicious flagging, retains data in memory for approximately one hour, and exposes it read-only via /dsh-egress/snapshot.json. The plugin does not initiate its own requests, write to disk, or store request bodies or Authorization headers — it pass-through wraps the egress. Install from the plugin marketplace or run install_bundle against a local directory. Requires the webServer and settings slots. MIT 0.2.0.
Read MoreChronic Disease Assistant: Recording Hypertension and Diabetes Indicators Locally
Chronic Disease Assistant is an Agent-oriented personal chronic disease management Skill, located at @user_a16b2e76/manxingbing, version 1.0.0, sourced from the community and categorized under life-service. It consolidates and tracks core indicators for seven chronic conditions—hypertension, diabetes, hyperlipidemia, cardiovascular/cerebrovascular disease, COPD, chronic kidney disease (CKD), and obesity/metabolic syndrome—including blood pressure, blood glucose, lipid test reports, weight/BMI/waist circumference, SpO2, peak flow (PEF), kidney function (eGFR/creatinine), medication adherence, symptoms, exercise, and diet. It automatically generates statistics and comprehensive reports, supporting export via WeChat long image, WeChat plain-text summary, Tencent Docs Markdown, Word .docx, and printable PDF. It runs entirely locally with zero required dependencies (Python standard library only; .docx export optionally requires python-docx) and needs no API key, making it suitable for sensitive health data. This article introduces its features, installation coordinates, typical trigger scenarios, and usage notes.
Read MoreAmap Ride-hailing Skill: Zero-Configuration Invocation of Ride-hailing and Route Planning
Amap Ride-Hailing is a community-edition Skill on SkillHub (v1.0.3; category: life-service; publisher: user_b95ee7e5). It one-tap launches ride-hailing in the Amap app, works out of the box with zero configuration, and includes driving/transit/walking/cycling route planning (coordinate version + address version), IP-based geolocation, nearby search, POI search, and POI details. The official description states it can be used without applying for an API key, but the platform tag shows an inconsistency (requires_api_key=true), which should be confirmed before installation. As of the data snapshot, its rating is 100000, downloads 872, and installations 0. This article introduces its positioning, features, installation entry, typical usage, and applicable limitations.
Read MoreJigang · Meeting Minutes and Resolution Follow-up: Turning Meeting Resolutions into Action Items That Never Fall Through the Cracks
This introduction focuses on "Jigang · Meeting Minutes & Resolution Tracking" (@user_bbfb885f/jigang-meeting-action, v1.3.1, community source, office-efficiency category). It extracts meeting elements and resolutions from meeting records, transcripts, Tencent Meeting intelligent summaries, rough notes, and chat logs. It verifies items—specifically Who, What, When, and Status—row by row, and outputs ready-to-send supervision documents, resolution tables for OA/DingTalk/Feishu/WeCom, ICS/CSV/JSON files, and standard meeting minutes. The core feature is "Responsibility Vacuum Verification V1–V8," which explicitly lists verifiable issues such as V1 (resolution without an owner), V3 (soft deadlines), V4 (responsible person not in the attendance list), and V8 (ambiguous references). If no resolutions can be extracted, it explicitly reports an error and provides three solutions, ensuring it never "succeeds silently." The design is strictly offline, read-only, and offline-networked; it does not read sensitive credentials, though `pypdf` is required for PDF extraction. No API key is required for activation, and it falls under the alipay paid category. At the time of writing, the official page was inaccessible, and installation commands and configuration fields were missing from verifiable sources; therefore, these details are marked as "pending verification" without speculation.
Read Moredsh-home-monitor: Self-hosted read-only health checks with a single config.json
lowspecai's dsh-home-monitor targets DSH 0.2.0-rc.2+ and utilizes server/config.json to centrally manage HTTP, TCP, SSL, and optional read-only Docker detection. The same logic can be registered as five DSH monitor_* tools or function as a standalone MCP stdio service. The article details prerequisites, dsh-tools symlinks, profile patch integration, configuration fields, and MCP registration examples, while emphasizing security considerations such as read-only access, no outbound filtering, and Docker being disabled by default.
Read MoreWanxiang Master Agent Clone Designer: Transforming WeChat Customer Service into Operationally Managed Digital Clones
This article introduces the community Skill "Wanxiang Master Agent Split Designer." Targeting WeChat Official Accounts and Enterprise WeChat Customer Service scenarios, it is based on the universal SKILL.md format and maintained by user_71e1dad5 (Version 0.5.1, Category: ai-agent). It consolidates personality, knowledge bases, FAQs, memory, hotwords, and role images into a single maintainable avatar configuration, covering four actions: creation, design, configuration, and optimization. It requires no additional connectors, but users must provide their own API Keys. The article outlines its positioning, capabilities, the method for activation via SkillHub, suitable and unsuitable scenarios, and precautions. Due to the official page being unavailable at the time of writing, unverified installation commands and configuration fields have been omitted; users are advised to defer to the official page for accurate information.
Read MoreOption Volatility: Screening Option Strategies Using the Dual Signal of IV Percentile and IV-HV
This article introduces the community Skill "Options Volatility" (coordinates @user_09427764/querylucky123py, version 1.0.0, category professional, source community, publisher deleted_user_635624, page-labeled requires_api_key=false). It is a QueryLucky-produced, all-asset options strategy scanning engine: inputs are real-time options chains and futures daily candlesticks; outputs are Top-3 high-win-rate strategy recommendations for each of 20 mainstream instruments, including leg details, risk control, Greeks, and win-rate scores. Core capabilities include dual-signal determination via IV historical percentile and the IV-HV20 spread; any conflict results in a neutral rating, avoiding the erroneous issuance of sell-side strategies when the percentile is low but IV is high. Strategy-level thresholds apply: a short straddle requires IV-HV ≥ 3pp, a short strangle requires ≥ 1.5pp, and an iron condor requires only a percentile ≥ 50. The组合保证金 whitelist is limited to au/rb/i/p/a/b for generating vertical/ratio/diagonal spreads, with SHFE-listed products automatically blocked. For short straddles/strangles, the maximum loss is displayed as infinite, not as a fictitious amount. No installation commands are provided on the official page or via the API; this article has not verified segmented instructions for Cursor, Codex CLI, or Claude Code, and thus includes no unverified commands. Input field names, return schema, and calling examples should also be based on the official page.
Read MoreDSH-profile-stats: Read Local Session Token and Usage Statistics in the DSH Sidebar
dsh-profile-stats is a DSH client plugin maintained by Lovenndme. It displays cumulative and daily token counts, 14-day sparklines, active streaks, 12-month activity heatmaps, usage stacked by model, cache composition (Cache Lens), and billing breakdowns by provider in the Web UI sidebar, utilizing data from local session logs. It requires DSH >= 0.2.1-rc.5 and < 0.3.0. It can be installed via Plugins → Discover (search) or by running `plugin_manager install_bundle`. MIT License, Version 2.4.12.
Read Moredsh-websearch-keyless: Enable DSH Native Web Search Without API Key
Introducing lovehin's DSH plugin dsh-websearch-keyless: When DeepSeek API Key is unavailable causing web_search failures, it sets searchProvider to opencode-keyless via bundle, sequentially degrades to call keyless endpoints of Tavily, TinyFish, and Exa, and only takes over web_search. Includes official npx installation commands, dump-config verification, Gotchas, and troubleshooting table. MIT license, version 0.1.0, Node >=20.3, tested with DSH 0.2.1-rc.3 in README environment.
Read MoreShenxiao Office Assistant Consolidates Search, Official Documents, and PowerPoint into a Unified Access Point
Deep Know Office Assistant is a comprehensive Agent Skill developed by Deep Know Trusted Intelligence, a subsidiary of Beijing Caizhi Technology. It integrates authoritative retrieval, formal document drafting, and presentation creation into a single interface. By automatically routing capabilities based on the user's deliverable requirements, it eliminates the need for manual module selection. The system allows for the retrieval of original policy regulations and administrative guidelines, facilitates city-specific comparisons, and verifies subsidy eligibility, with every conclusion anchored to authoritative source texts. It supports the drafting of over 20 document types—including notifications, requests for instructions, reports, official letters, and meeting minutes—delivered primarily as Word files. For statutory official documents, it can generate "red-header" official versions, while for long-form research reports, it prioritizes the creation of a factual working draft before proceeding. The presentation capability generates native PowerPoint (.pptx) files where every shape and table remains fully editable. It supports the direct conversion of Word materials into PPT format and includes five built-in style presets and six canvas specifications. All factual materials are sourced from authoritative document libraries rather than open web searches, and citations include superscript markers for easy verification against original texts. Prior to delivery, the content undergoes AI self-validation accompanied by a traceability verification report for item-by-item review. The service is available immediately via the WorkBuddy Expert Marketplace or by connecting to the Deep Know Trusted Workbench to obtain an access key without a phone number. New registrations include a bonus of 100,000 credits, with an additional 100,000 credits awarded upon real-name authentication.
Read MoreAmazon Review Sentiment Analysis · Positive/Negative Tendencies: Conclusions by Dimension, Not Just Overall Scores
This community Skill (@user_507a5056/review-sentiment, v1.5.2, classification: data-analysis) addresses the pain point in e-commerce data analysis of relying solely on aggregate review scores: different attributes of the same product often yield polarized word-of-mouth, and overall scores flatten dimension-level information, leaving operations teams unable to pinpoint which specific aspect to improve. It first determines the sentiment of individual reviews, then aggregates by thematic dimension, producing positive and negative conclusions broken down by dimension. The same package provides ARI's full review analysis capabilities: subscribing to ASINs and collecting reviews, browsing and filtering reviews, viewing star ratings/keywords/trends/radar charts, generating VOC and deep-insight reports, a product operations workflow, watch-based product change monitoring, negative review alerts and a negative review workbench, industry benchmarking and category rankings, as well as exporting reviews to CSV and reports to Markdown/HTML. Trigger words cover sentiment analysis, good vs. bad review comparison, word-of-mouth tendency, and positive/negative classification in both Chinese and English. A hard prerequisite is an ARI API key (ari_live_*). At the time this article was written, the official page failed to be fetched twice; therefore, installation commands, configuration fields, and parameter details have not been transcribed—only verifiable content such as coordinates, key requirements, and the feature list is retained. Please verify the official page before integrating.
Read Moredsh-projectflow: Per-project Persistent Context and Task State
DSH-ProjectFlow, maintained by lord-of-vaegirs, is a DSH memory plugin that stores conventions, decisions, progress, references, and task checkpoints in local Markdown/JSON files under a stable `projectId` and the `$DSH_HOME/projects/` directory, decoupling them from the session working directory. It supports the `/project` command, fork/merge, reference metadata without duplicating content, and a Project workspaces panel on the web side. Installation is available via release tarball, npm, or GitHub source, requiring DSH >= 0.2.1-rc.2. Based on the official README, this article introduces installation, quick start, and usage notes.
Read MoreAudit Plan Strategy Possibilities: Generator, transforming annual ASM documents from manual assembly into a single workflow
The Audit Strategy Memorandum (ASM) Generator is a community skill on SkillHub (@user_196a272e/asm-generator, v1.0.0, categorized under "professional") designed to generate Annual Audit Strategy Memorandums. Its trigger words include ASM, 审计策略备忘录 (Audit Strategy Memo), 审计策略 (Audit Strategy), and Audit Strategy Memo. For listed companies, users need only provide the company name, and the skill autonomously collects public information; for non-listed companies, user-provided Excel financial statements are required. This article outlines the tool's positioning, verified core functions, coordinate specifications, "no API Key required" annotations, reproducible trigger methods, as well as its target audience and limitations: it is a document generation tool whose output requires review by auditors and does not replace professional judgment; the scenario for non-listed companies depends on the quality of user-provided reports. As a community-sourced project in its early stages, its current installation count is 0, with 901 downloads.
Read MoreSkill for Generating Equity Investment Proposals: Organizing Raw Semiconductor Project Data into Standardized Investment Recommendations
The Equity Investment Recommendation Generation Skill (@user_238915fc/semiconductor-investment-report) is a tool designed for the semiconductor industry to generate investment recommendations. Maintained and published by SkillHub user user_238915fc, currently at version 1.8.7, it originates from the 'community' source, falls under the 'professional' category, and requires no API key. Leveraging raw project materials submitted by users (such as Business Plans, White Papers, Due Diligence Summaries, Financial Data, and Team Resumes), the skill performs information extraction, structured classification, and professional in-depth interpretation. It strictly adheres to a mandatory 12-chapter template (featuring tabular formatting and 1.1/1.2 sub-section structures) to output industrial investment recommendations that are standardized in format, traceable in information, and profound in argumentation. Enforcing zero fabrication throughout the process, missing modules are uniformly marked as "Data Missing," and contradictory modules are marked as "Data Conflict." It supports one-click export to DOCX (official layout), PPTX (presentation), and Web (HTML) formats. Additionally, the skill functions as a linked component with Due Diligence (DD) skills; it can generate the final recommendation report based on DD, EV, and FA results combined with original materials after the due diligence phase is completed. This article covers its capabilities, installation and activation methods, typical usage scenarios, and application precautions.
Read Moredsh-workspace-sidebar: Sidebar Workspace Overrides and Conversation Fork Tree
dsh-workspace-sidebar is a DSH Web client plugin maintained by longyi-0x0 that completely replaces the sidebar `ui-workspace`. Based on the official `@deepseek-ai/dsh-client-ui-workspace` 0.2.1-rc.5 source code, it implements a forked conversation tree, branch folding, archiving all sessions, and style hot-swapping. To install, run `dsh plugin --profile desktop add github:longyi-0x0/dsh-workspace-sidebar`; a DSH restart is required after installation. This plugin is designed exclusively for the Web client sidebar and must not be used concurrently with the official `ui-workspace` in the same profile.
Read More