Financial Statement Analysis Expert: Interpreting the Three Statements as a Diagnostic Report
The SkillHub community skill "Financial Statement Analyzer" (Coordinate: @user_053d27d0/wzx-04, Version 1.0.0, Category: professional, Source: community) utilizes a three-layer funnel diagnostic engine to read the three major corporate financial statements and outputs structured HTML reports featuring Chart.js visualizations, risk scoring, and prioritized management recommendations. It supports six input formats: .xlsx, .xls, .csv, .pdf, .png, .jpg, and .docx. Tagged with `requires_api_key=false`, it requires no additional API keys. The article is organized into six sections: Positioning, Core Features, Installation & Activation, Typical Usage, Applicable Scenarios & Precautions, and Conclusion. The installation section accurately states that no independent command has been officially provided, the client references the skill by coordinates, and there are no differentiated instructions for Cursor, Codex CLI, or Claude Code; no inferences are made. The usage section covers triggering via file uploads, `/analyze`, or `@Financial Statement Analyzer`, along with keywords such as "analyze statements," "cross-checking," "M-Score," and "DuPont analysis." The precautions emphasize that the report does not replace audit procedures, image recognition errors may propagate to the scoring, and page ratings and download counts represent internal site popularity records.
Read MoreLegal Opinion Skill: Drafting Legal Arguments in the Style of a Formal Opinion
A Skill designed for legal argumentation scenarios, identified as `@indiv-qf/qf-legal-opinion-v2`, published on the SkillHub community under the "professional" category. The current version is 2.9.0, released by `user_eb2234bc`. It bundles legal opinion drafting, legal risk assessment, compliance analysis, argumentation on state-owned assets regulatory matters, M&A/disposal project legal argumentation, and decision-making review materials into a single call. Utilizing trigger words and case background, it outputs text organized according to the formal opinion style, covering the common structure of opinions: factual background, legal analysis, and conclusions. Official Chinese trigger words include "write legal opinion," "issue legal opinion," "compliance analysis," "legal consulting opinion," and "risk assessment report," while English trigger words include "legal opinion," "legal risk assessment," and "compliance opinion." The invocation uses Alipay AI pay-as-you-go pricing at 0.20 CNY per use, with a policy of no delivery without payment. The tags indicate that no separate API Key configuration is required. As the retrieved material lacks specific installation commands or differences between Cursor, Codex CLI, and Claude Code, the body text will exclude unverified commands and only provide enablement methods and typical invocation examples. It is important to note the boundaries: it is not suitable for simple legal consultation Q&A, contract drafting, or legal document translation, as these fall outside its design scope. The official URL is https://api.skillhub.cn/indiv-qf/qf-legal-opinion-v2.
Read MoreKizuna: Agent Directory and Reliable Delivery in the DSH Ecosystem
Kizuna (01men/kizuna) provides Agent directory, Hub relay, and edge-side mailbox for DSH workflows, emphasizing acknowledgments, retry gradients, and append-only audit trails. The article explains the division of responsibilities among kizuna-hub / kizuna-box / DSH plugin, the official installation command `dsh plugin add github:01men/kizuna`, as well as typical usage patterns of Hub API, A2A passthrough, and MCP, along with self-hosting considerations.
Read MoreDingTalk Meeting Notes, Summaries, and Action Item Extraction: Breaking Down a Meeting into Actionable Plans and Checkpoints
This is a community Skill — "DingTalk Meeting Transcription, Summary & To-Do Extraction" (credited to Jianshi AI, available at @user_fccf428b/china-business-dingtalk-minutes, version 1.0.0, categorized under office-efficiency; requires_api_key=false). Targeting operation and collaboration teams in Chinese SMEs, it delivers DingTalk meeting transcription, summarization, and to-do extraction completed within business decision-making, project collaboration, meetings, documentation, or organizational management scenarios, and produces actionable solutions, checkpoints, and risk boundaries. The article outlines the capability scope, the official entry point, and a five-step reproduction workflow, and also notes items that require independent verification: the provided materials do not include installation commands, differences across IDEs, accuracy rates, or maximum duration limits.
Read Moredsh-problem-bank: Structured Math Problem Bank and Variant Question Generation
dsh-problem-bank is a DSH workflow plugin (v0.6.2, MIT) by maintainer ziyans2022, designed for building math question banks and generating problem variants. Problems are stored on disk as markdown with frontmatter, and the plugin works together with a bilayer knowledge-point graph, a solution-technique library, multiple solution methods, a six-dimensional difficulty profile, and a variant quality gate. It assists with generating problems that share the same technique but differ in problem statement, performing near-duplicate checks, and assembling papers under constraints. It has not yet been published to npm and can be installed via git with a full commit or from a local directory. The data directory bankRoot is recommended to use an absolute path, since the plugin writes to disk directly without going through a sandbox. Machine-based review does not replace manual problem verification.
Read MoreBull Stock King Dragon-Tiger List: Analysis of Capital and Trading Seats on the A-Share Dragon-Tiger List
"Niuguwang Dragon-Tiger List" is a Dragon-Tiger List analysis Skill for A-shares (version 1.0.2), published by user_f884a472 and sourced from community, classified as "professional" in SkillHub. It requires no API Key. Its core capability is to display the nature of capital sources (institutional investors/hot money, etc.) and the attributes of trading seats for stocks that appear on the list each day, helping users identify the operators behind unusual price movements and to serve as a capital-market reference for next-day trading decisions. The article explains in sequence its positioning, core capabilities, installation and activation, typical usage, and applicable limitations. It should be noted that the official page and the API were inaccessible at the time of writing, and the provided materials do not include installation commands or invocation examples. This has been faithfully noted in the article, with a recommendation to refer to the official documentation for accurate instructions. Page statistics show a rating of 100,000, 1,296 downloads, and 0 installations, so actual adoption levels remain to be validated.
Read MoreHuang Yuanyu's Classic Prescription TCM Qi Movement Pattern Diagnosis Skill: Turning One-Qi Cyclical Flow into a Reasoning-based Diagnostic Tool
huangyuanyu is an open-source Traditional Chinese Medicine (TCM) professional skill on SkillHub, located at @user_ff4d9420/huangyuanyu, with version 2.9.0, categorized under education, and published by user_ff4d9420. Centered on *Sheng Yuan Xin Yuan* (The Heart of the Four Sages), it integrates eleven original works by Huang Yuanyu, such as *Su Ling Wei Yun* and *Shang Han Xuan Jie*. By digitizing the "Circulation of One Qi" and the "Earth as the Pivot for Four Images" systems, it constructs a diagnostic reasoning framework featuring the Central Qi as the axis and the Liver, Heart, Lung, and Kidney as the four wheels. It features built-in six-meridian Qi transformation deductions, standardized diagnostic decision trees, Huang's formulas and classical compatibility, a three-dimensional retrieval index for pathogenesis, symptoms, and formulas, as well as warnings against mistreatment such as "bitter-cold damaging the center" and "cloying hindering the spleen." By inputting physical complaints, it derives paths of abnormal Qi ascent and descent, locates the root of Central Earth decay, and matches therapeutic principles and formulas. It can also be linked with the Li Ke Six-Meridian Skill and the Wu Jutong Warm Disease Skill for comparative analysis. Trigger words include Huang Yuanyu, Circulation of One Qi, Ascending on the Left and Descending on the Right, Earth Pivot and Four Images, Central Qi as Axis, Water Cold and Earth Dampness, and Louqiuzi. It is not applicable to pure external excess heat syndromes, true Yin deficiency with fire effulgence, Warm Disease Wei-Qi-Ying-Xue differentiation, or pure Yangming Fu-excess heat syndromes. It is intended solely for academic study of TCM theory and does not replace offline diagnosis and treatment by licensed TCM practitioners.
Read Moredsh-semantic-recall: Retrieve Original Messages from Old Conversations by Semantic Search Even When Starting a New Conversation
dsh-semantic-recall is a DSH memory plugin that建立 (builds) a semantic index by reading the last N days of sessions locally in read-only mode. It provides four agent tools, including `session_recall`, and features a built-in settings page. It supports fuzzy retrieval across workspaces, asks clarifying questions when relevance is insufficient, and ensures multi-source workspace isolation. It can also be combined with long-term memory plugins to drill down into the original text. Installation is performed via `plugin_manager`'s `install_bundle` with the target set to `dsh-semantic-recall`; it requires Node >= 22 and DSH >= 0.2.1-rc.5.
Read Moredsh-council: Turning DSH Sessions into Concurrent "Group Chats"
@zelystaric/dsh-council is a DSH workflow plugin that orchestrates a session into director-led decomposition: 8 members with customizable roles execute concurrently on frameworks such as codex/claude/deepseek/image, followed by reviewer preliminary review and director final review. The group chat page displays stages and elapsed time in real time. It supports official or relay APIs, and strictDelegate blocks parent-session tool writes. Based on the GitHub README and package.json (0.2.17, MIT), this article explains the installation, configuration, and first-use sequence, and is suitable for DSH users who need multi-agent collaboration with observable processes.
Read MoreMulti-Agent Collaboration Framework: Encapsulating Multi-Agent Division-of-Labor Workflows into Skills
The multi-agent collaboration framework is a prompt-based Skill built on SKILL.md, maintained by community publisher user_2fd890c9. It utilizes prompt templates to standardize the multi-Agent division of labor workflow for the "Architect-Craftsman-Verifier" roles, generating structured JSON. It supports two modes—deep reasoning chains and role-based collaboration—and relies on the OpenAI API key and httpx, without official functions or code execution.
Read Moregame-art-rebuilder：One-Click Reconstruction and Repainting of Game Art Visuals
game-art-rebuilder is a generative Skill designed for game art, located at @user_aecc9bb1/game-art-rebuilder. It is community-sourced, categorized under design-media, version 1.0.1, and tagged with requires_api_key=false. Starting from one or more reference images, video frames, links, or named works, it remakes or redesigns static game art, covering types such as CG, KV, splash art, concept design, storyboards, VFX frames, UI screenshots, and posters. It supports three granularities: high-fidelity reconstruction, structural redesign, and low-reference free redesign. It can assign roles to multiple references, maintain storyboard continuity, handle layered KV layouts, and perform complete Chinese visual analysis. By default, it generates images directly, but can also produce GPT Image prompts as needed. The core philosophy is to derive similarity from controllable visual relationships rather than vague style vocabulary.
Read Moredsh-image-drag: Ctrl-Drag Panning for Image and PDF Preview in the DSH Right Sidebar
dsh-image-drag is a DSH client extension maintained by YuyuYuu-wyt. In the right sidebar, it enables Ctrl+left-click panning for previewing zoomed images, PDFs, and Office documents converted to PDF, and it injects a close button into each docked panel. It does not override native zoom or PDF text selection. Requires DSH >= 0.2.0-rc.2; it can be installed via npm or GitHub into the desktop profile. It is implemented entirely on the browser side, with no networking and no persistence. This article summarizes installation steps, usage, and known limitations based on the repository README and package.json.
Read MoreShuoruan · Listing Copywriting & Keyword Embedding Assistant: Generating Compliant Copy Across Five Platforms from Single Parameters
A listing copywriting skill designed for cross-border e-commerce sellers and operators. It accepts product parameters as input to generate compliant copy and 4-tier keyword embedding tables for five platforms: Amazon, Temu, Shein, TikTok Shop, and Shopify. Before export, it performs mechanical validation to check for character limit exceedances, forbidden words, and incomplete keyword coverage, blocking the export if any issues are found. It operates as a purely local generation tool and requires no API Key.
Read Moredsh-rule-guard: Rule and Target Verification Before Side-Effect Execution
dsh-rule-guard is a DSH admin-security plugin maintained by yunling-bionic (v0.4.3, MIT). It hooks into tools/pre-execute to perform fail-closed review of tool calls with side effects: by combining AGENTS.md, .dsh/rules.md, the data directory's RULES.md, and the user's most recent instructions in the session, it determines whether an action falls within the current objective; otherwise, it rejects the action and returns corrective suggestions. Submitted for review are operations including disk writes, non-read-only shell commands, network access, and sub-agent invocations; read-only operations within the workspace are exempt. The installation command is dsh plugin add dsh-rule-guard. Vague instructions will be rejected; this is a soft constraint rather than a security boundary.
Read MoreDeep Market Research Analysis Expert: Turning Market Research into a Reproducible Pipeline
To address issues such as unstable model research results, untraceable sources, duplicate information being treated as multi-source evidence, and the mixing of stale news with unverifiable content, this article introduces the community Skill "Deep Market Research Analysis Expert" @user_1e758512/deep-market-research (Version 2.3.4, classification professional). It decomposes research into a deterministic pipeline spanning Steps 0 to 8: intent routing, retrieval, deduplication, scoring, adversarial auditing, and curation. It employs NATO Level 4 source grading to annotate sources, requires cross-validation from at least two sources, and actively performs four filtering stages—deduplication, de-staleness, de-forgery, and de-contradiction—ultimately attaching confidence labels to the results. Platform-agnostic and zero-dependency by default, it utilizes only the LLM's built-in web_search/fetch and free REST APIs (requires_api_key=false). It supports optional MCP/API backends and degrades gracefully if the backend is unavailable. Outputs are generated in both HTML and Markdown formats, available in bilingual Chinese and English. It is applicable for market research, competitive analysis, industry landscape mapping, technology trend analysis, due diligence, and investment opportunity discovery. As official documentation does not provide specific installation commands, enablement steps for various environments are not speculated upon here.
Read MoreInvestment Banking Case Research Assistant: Delegating A-Share IPO Case Retrieval and Regulatory Logic Induction to Skill
The Investment Banking Case Study Assistant (`@user_44747f29/ib-case-study`) is a community-maintained "Professional" category Skill (version 2.0.11). Integrated with the IMA "IPO Practice Case Library (Continuously Updated)," it is positioned as a case study assistant for A-share IPOs, M&A restructuring, and refinancing, with capabilities focused on case retrieval, case comparison, and the summarization of regulatory logic. Trigger words cover: case study, comparable cases, review cases, inquiry cases, case comparison, regulatory logic, practice patterns, case summarization, case reference, and terminated cases. This article outlines its positioning, core functions, and verifiable information (including coordinates, classification, version, source, and the fact that no API Key is required). It provides typical use cases based on trigger words—such as identifying comparable cases, conducting case comparisons, summarizing regulatory logic, and reviewing terminated cases. Furthermore, it specifies the application boundaries: the case library is primarily focused on A-shares; platform ratings and download counts do not equate to the library's coverage; official installation commands are not provided in the materials; and users must confirm the installation channel and verify the returned results using real cases prior to integration.
Read Moredsh-plugin-updater: Check for Updates and Enable One-Click Updates on the DSH Plugins Page
The official plugin marketplace lacks an upgrade operation, so installed plugins often need to be manually uninstalled and reinstalled. dsh-plugin-updater adds an "Update" tab to the plugin page, checks for the latest versions from registries, Git, and local link sources, supports switching installation sources and performing compatibility prechecks, reuses the official removeBundle/installBundle pipeline for one-click upgrades, and provides options to open the local directory or a repository link. Installation can be done by adding the plugin and pasting the GitHub repository, or by running dsh plugin add github:YukiSuoug/dsh-plugin-updater.
Read Moredsh-skills: Managing Agent Skills in the DSH Sidebar
The DSH default discovers Agent Skills from paths like .dsh/skills and .agents/skills but lacks a centralized management interface. The community plugin dsh-skills (yixin2021, v0.1.0, MIT) adds a Skills panel to the left navigation: lists all root directory Skills, allows previewing SKILL.md files, enabling/disabling by moving to .disabled directory, batch installation from local folders or GitHub repositories, safe uninstall to .trash folder with restore capability, and supports adding external Skill directories with known Agent repository integration. Installation command: github:yixin2021/dsh-skills, CLI example: dsh plugin --profile desktop add (requires reload after installation). The plugin operates with the same file and git access permissions as the DSH process—review source code before installation.
Read MoreCAD Universal Converter: Cloud Conversion of DWG/DXF Drawings to PDF
CAD Universal Converter (Skill ID: @user_40969e1d/dwf-pdf-img-version, v1.0.0, community, design-media) is a Skill that batch-converts DWG/DXF drawings to PDF, PNG, or JPG, or converts between different DWG versions. It does not parse drawings itself; instead, it invokes the 52cad.com cloud service: first uploading the file to obtain an f_id, then polling get_status until completion, and finally downloading the result. Documentation shows requires_api_key=false, meaning no API key configuration is needed. The publisher is user_40969e1d. This article describes its features, working principle, installation entry point, typical usage, and caveats (cloud dependency, compliance requirements for sending files externally, format conversion only), and provides the official URL.
Read Moredsh-feishu: Running DSH Agents and Conversation Backend on Feishu / Lark
**dsh-feishu** is a community-maintained DSH (Deer Flow Shell) integration for Feishu/Lark (alpha preview), pinned to DSH **0.2.1-rc.5** and the Feishu SDK **1.76.0**. It provides persistent sessions per app/conversation/topic, attachment and group-topic support, workspace read/write and isolated shell access, plus optional **toolGroups** for the Feishu API and approval cards for requesters. There is no one-click npm package; you must build an archive on a Linux target such as Ubuntu 22.04 and deploy it via **install.py** and a user-level **systemd** service. It is suitable as a self-hosted foundation for Feishu agents; before installing, read the **SECURITY** policy and license, and be aware that workspace isolation is not per-user.
Read MoreLong-form Novel Generator: A Writing Assistance Skill Covering Settings, Outlines, Chapters, and Review
Community-sourced writing skill, located at @user_3cdc1868/greatnovels, version 1.0.0, classified as content-creation, following the general SKILL.md format, with the tag requires_api_key=false. It decomposes long-form writing into four stages—settings, outline, chapter writing, and proofreading—to avoid inconsistencies in characters and foreshadowing that can result from single-pass, whole-text generation. This article verifies facts against captured materials: the introduction only states the capability scope and does not provide installation commands, configuration fields, or official usage examples; therefore, the installation and usage sections only describe the general loading method and a reproducible call order, without fabricating details. The captured page data shows a score of 100000 and 1234 downloads, which are inconsistent with 98964 and 1146 in the task clues; the main text adopts the page values and explicitly notes the discrepancy, alerting readers that such statistics are subject to change and should not be used as selection criteria. Suitable for long-form content creation that requires multiple stages and individually verifiable deliverables. Precautions include: it is community-sourced and not officially maintained, versions may change, and you need to confirm that the target Agent supports loading the general SKILL.md beforehand.
Read MoreAgent Ops: Agent Runtime Visualization and Observability Dashboard in DSH
Agent Ops (dsh-agent-ops) is a DSH client plugin maintained by xinranmn-wq. It provides a full-screen HUD in the bottom-right corner of the Web UI, featuring information feeds such as the memory cloud, a 42-card rotating ring, and TASK FEED. It polls the local GET /dsh-agent-ops/state endpoint to display tool calls, tokens, subagents, and event logs; LIVE FEED and SIMULATION switch automatically, and data never leaves the local machine. Installation is available via Agent install_bundle, dsh plugin add link:local path, or Settings→Plugins; it requires shell.overlay and session/event. MIT 0.1.0.
Read MoreMega Bundle of Finance & Tax Expert Team Matrix: Instant Access to 57 Finance & Tax Experts and 13 Expert Groups
Tax & Finance Expert Team Matrix Assembly Bundle (@user_11064e10/tax-expert-team-matrix, version 3.18.64, source: community, category: professional) is a batch-assembly Skill: it installs 57 tax-and-finance domain experts and 13 cross-domain collaborative expert teams into “My Experts” in one go, and comes with team-lead triage, preset workflows, and invocation guidance, replacing the practice of manually creating each expert role one by one. The tag requires_api_key=false indicates that no separate API key configuration is required. This article verified the capability description and accurately notes that the official page does not provide an installation command, SkillHub CLI usage, or reproducible invocation examples; therefore, no installation command is provided and no examples are written. The itemized lists of the 57 experts and 13 expert teams likewise need to be confirmed on the official page. The page-scraped metrics are rating 100000, downloads 1220, and installs 0, which differ from the rating 99050 and downloads 1127 reported in the task clues; the page data is taken as authoritative. As this is professional-category content, tax-and-finance conclusions still require manual review.
Read MoreMedical Analysis of Chinese Patent Medicines: A Skill for Breaking Down Medical Affairs into Nine Executable Modules
This article introduces the "Traditional Chinese Medicine Medical Analysis" Skill (user_31815b66, professional, v1.0.0) on SkillHub. Designed to meet the needs of analyzing Traditional Chinese Medicine (TCM), specific varieties, or therapeutic areas from a Medical Affairs perspective, the Skill covers nine modules: Medical Insights & Intelligence, Medical Strategy & Syndrome-Disease Integration Positioning, EBM System & Grading (L1a-L5), Evidence Generation (RCT/RWE/PRO/IIT/Phase IV PASS, etc.), Medical Information & Education (Dual-track MSL), Pharmacoeconomics & Value Evidence, Integrated TCM-WM Pathways & Guideline Promotion, Safety & Label Management, and Pre-marketing Medical Interfaces. It features a built-in foundation.md module; allows switching between "Entry" and "Senior" tiers based on user proficiency; and strictly delineates exclusive medical functions, collaborative interfaces, and non-medical boundaries. It enforces three specific compliance constraints for TCM: anchoring to TCM terminology in the approved label, segregating scientific exchange from promotion, and ensuring safety data aligns with the latest labeling. Metadata tags include requires_api_key=false and pricing_type=alipay. The text explains installation and acquisition methods, provides typical trigger phrases for usage, and outlines applicable scenarios and limitations.
Read Moredsh-kon-fee-widget: DeepSeek Department Expense Ledger on Your Desk
dsh-kon-fee-widget is a DSH fun dress-up plugin maintained by xiao2236. Based on the original Qingyin tea table and the five-person seating poses, a single click pops up DeepSeek balance, today's spending, and peak/off-peak tags; turn/end accumulate daily usage and, at the end of each round, display the round's cost in a notification window. The host fetches the balance via /balance (API Key or DSH account) and computes Flash/Pro peak/off-peak pricing according to the official price list. It supports double-click Chinese/Japanese line easter eggs, character selection, and drag-to-snap to edges. Web install: dsh plugin --profile web add github:xiao2236/dsh-kon-fee-widget; configure ~/.dsh/kon-fee-widget.json. Code under MIT; character assets carry third-party notices and are unofficial fan works.
Read More