All-in-one Novel Creation: A Skill for Batch Generating Novel Chapters from a Knowledge Base
Novel Writing Full-Skills (or "Novel Creation All-Rounder") is a Skill designed to write and continue novel content based on a knowledge base of novel settings. It produces editable .md files, supports chapter modification, optional Word export, and one-click knowledge base import. It offers two modes: single-chapter and batch/continuous generation (generating an entire volume at once or generating continuously based on set specifications without chapter-by-chapter confirmation). Targeting genres such as sci-fi, mystery, and fantasy, the body text requires a minimum of 2,500 words, with single outputs ranging from 2,500 to 5,000 words, and word count statistics are provided with each chapter delivery. It is triggered via natural language commands such as "write novel," "continue novel," "write next chapter," "write plot based on outline," "batch generate chapters," "modify chapter," "export Word," or "import knowledge base." It is not applicable for news, academic papers, official documents, merely organizing settings, short jokes, or copywriting. The crawled data and task clues cited in this text show inconsistencies in scoring and download counts, which have been noted in the text. Installation commands and configuration fields could not be verified and have been left blank.
Read MoreWorkflow Decomposition Assistant: Breaking Down "What People Are Doing" into Atomic Tasks
Workflow Breakdown Assistant is a job/workflow decomposition Skill for HR, OD, and business managers, located at @user_dbd0e202/work-breakdown-assistant, version 1.0.0, published by user_dbd0e202 (source: community), category business-ops, tag requires_api_key=false. It systematically decomposes an individual's or a team's actual work into atomic tasks, identifies the core skills required for each task, and produces six outputs: a task list, a workflow list, a dynamic job profile, a job skills map, a task-to-metrics mapping, and diagnostic insights. It is suitable for scenarios such as breaking down job responsibilities or work content, analyzing team workflows, organizing the composition of job tasks, conducting work analysis or job analysis, generating job task lists or job descriptions, identifying core skills and competency elements, assessing the relationship between tasks and performance metrics, generating employee interview outlines, and analyzing interview notes to extract work tasks. This Skill focuses only on clarifying what people do and what capabilities are needed, and does not provide suggestions for division of labor between humans and machines or for AI replacement. Official materials do not provide installation commands or configuration fields; before use, confirm how to load Skills in your environment. Public access is via the SkillHub page and the official API address.
Read Moredsh-prompt-sharpen: Pre-send inspection, only fill missing items
dsh-prompt-sharpen is a DSH plugin maintained by MeloFang. Before sending, it performs local zero-token rule checks to determine if the prompt is missing information across four categories: output format, constraints, acceptance criteria, and context. The Composer button, /sharpen, or sharpen_prompt only supplements missing items and allows for one-click restoration of the original text, whereas checkup_prompt is for diagnostics only. It supports installation via web profile and configurations for providers and models, and is open-source under the MIT license.
Read MoreCorporate Verification Skill: Multi-Company Implicit Association Mining and HTML Due Diligence Report
This article introduces the "Company Verification" Agent Skill on SkillHub (`@user_718560f7/company-intel-agent-sige`, v1.2.0, business-ops), designed for enterprise due diligence and multi-company relationship verification. It generates a structured profile for a single company using four data sources: business registration, equity penetration, judicial litigation, and bidding information. When explicit signals such as common shareholders or overlapping positions are absent, it cross-validates implicit signals—including contact information, registered addresses, historical personnel overlaps, nominee shareholding, and shell company indicators—to identify relationships. Based on the China Securities Regulatory Commission (CSRC) standards for affiliation, it provides a qualitative assessment and outputs a self-contained HTML report. Full capabilities are unlocked by connecting to either Tianyancha or Qichacha, as well as PKULaw. The article outlines core features, verifiable official addresses and version information, necessary client and data source permissions to confirm prior to installation, applicable scenarios, and limitations; it does not fabricate installation commands or case data.
Read Moredsh-plugin-paper-radar: Medical and Life Sciences Literature Search and OA PDF Retrieval
melano514/dsh-plugin-paper-radar registers the paper-radar skill for DeepSeek Harness, retrieves literature through Europe PMC, PubMed, OpenAlex, Crossref, uses Unpaywall et al. to determine OA status and validate PDF download, and on failure generates a decision card containing four channels: 科研通 (ScienceTalk), Sci-Hub, SciDown, and X-MOL. Written in pure Python standard library, MIT licensed, installation command `dsh plugin add melano514/dsh-plugin-paper-radar`, requires Python 3.8+ and Node >=20.
Read MoreLaozi Asks: Turning the Dao De Jing into a Callable Decision-Making Framework
"Laozi Wenda" is a community-sourced Skill in the education category with the ID @user_1876b8a1/laozi-wen-dao, currently at version 1.0.0. Based on the 81 chapters of the *Tao Te Ching*, it provides four types of content: original text interpretation, core frameworks, modern application derivations, and quick-reference decision tables. It leverages frameworks such as "non-action (wu wei)," "embracing softness," "non-contention," and "the movement of the Dao is to return" to assist in decision-making, self-cultivation, and spiritual enlightenment. The article outlines its positioning, core functions, typical use cases, applicable scenarios, and caveats. It also notes that the official documentation does not provide specific installation commands or reproducible invocation examples; therefore, users should refer to the official page for accurate information.
Read Moredsh-overleaf-sync: Bidirectional Sync between Local LaTeX and Overleaf in DSH
dsh-overleaf-sync is a DSH networking tool plugin maintained by measureer. It binds local LaTeX directories to cloud projects through the Overleaf collaboration channel (non-Git bridging), providing pull/push/diff/watch operations along with conflict resolution tools. It can automatically run git init after pulling. The plugin supports Cookie-based login, defaults to push dry-run mode, stores credentials in ~/.dsh/overleaf-sync/, and includes a settings page and a right-sidebar GUI. Installation requires adding a Release tgz file or linking a local path to the profile. The typical workflow is login → link → pull → diff → push.
Read Moredsh-staging-guard: Making the "Results Area" Read-only with Pre-execute Hooks
dsh-staging-guard is a DSH admin-security plugin released by mdl233. It intercepts write/edit/str_replace_editor operations targeting the output area under protected roots at the tools/pre-execute stage, guiding the agent to only modify the draft area. Protected roots are determined by the union of roots.json and the session cwd. Finalization requires running promote.mjs outside the host. It supports environment variable configuration, disabled emergency pass-through, and test-scope.mjs offline testing. The author clarifies that it serves as a soft boundary to prevent accidental modifications rather than a security boundary. Installation can be done via node install.mjs or dsh plugin add dsh-staging-guard. MIT 1.0.0.
Read Moredsh-agent-status: Track Agent progress in the DSH session title bar
dsh-agent-status is a DSH client plugin maintained by mdktelmdktel. It displays a real-time capsule in the session title bar showing the current action, step N, elapsed time for this round, and TODO x/y. Starting from 0.2.0, on Windows it can provide a top-right corner, always-on-top HUD via desktop-hud.ps1. Data is fetched from GET /agent-status/api/state after the host folds session events; the client polls approximately every 1.2 seconds. The install bundle is github:mdktelmdktel/dsh-agent-status. After adding the dependency and bundles in your profile, run pnpm install and refresh the page. Zero configuration, MIT license. Note: if you open a session midway, the step count does not backtrack; the API is loopback-only and has no authentication.
Read Moredsh-daily-luck: DSH Web UI Tarot Daily Fortune (5D Scoring + Q&A)
The DSH Web plugin dsh-daily-luck ("Today's Good Luck Arrives") by Maxim-Kerman-Lin: After entering a birthday, it generates a hidden base fortune locally in the browser using an approximate ephemeris, shuffles and lets the user select a Major Arcana card, and overlays it to produce five-dimensional scores for career, love, wealth, physical and mental well-being, and social interaction, along with a comprehensive commentary. It optionally calls the current default model via the Host endpoint `/dsh-daily-luck/analyze` for lightweight analysis and Q&A, falling back to local interpretation upon failure. Installation: `dsh plugin add github:Maxim-Kerman-Lin/dsh-daily-luck`. Data is stored in localStorage. MIT 1.2.0.
Read MoreStock Long-Short Factor Evaluator: A 13-Factor Long-Short Review of a Specific Stock
This is a bull/bear stock evaluation Skill released by community developer user_1832f53b (located at @user_1832f53b/stock-bull-bear-evaluator). Version 1.1.0. Sourced from the community and categorized as professional, it requires no API key. By entering a company or stock name in natural language (e.g., "Evaluate the bull and bear potential of BYD"), users can conduct a systematic 13-factor bull/bear assessment of A-share and Hong Kong stock targets. For bearish factors, the Skill proceeds to identify the company's mitigation measures and define quantifiable tracking indicators, then retrieves the latest progress via web search, and provides ratings categorized as "better than expected," "in line with expectations," or "worse than expected." The article outlines the core capabilities and limitations, official metadata, and enabling prerequisites (requiring host support for SKILL.md loading and internet connectivity), as well as reproducible trigger language examples, applicable scope, and precautions, including that factor definitions are subject to SKILL.md and that mitigation measures are analytical outputs requiring verification against official announcements.
Read MoreMarketing Calendar: Skill for checking marketing nodes and generating Template Marketplace search links
The Marketing Calendar is a Skill developed by YiQiXiu (Version 1.0.1, Category: business-ops) that requires no API Key. It queries holidays from the YiQiXiu Marketing Calendar and generates search URLs for the Template Store. It addresses two high-frequency, trivial tasks: checking marketing nodes (determining holiday schedules and coordinating warm-up and launch rhythms) and finding templates (obtaining direct search links to the Template Store to eliminate manual searching). It is suitable for operational automation workflows involving batch promotion planning and advance material preparation based on specific marketing nodes. This article covers its positioning, two verified core capabilities, and reproducible typical use cases (checking nodes, generating search links, and combining the two steps). Note: The official page could not be accessed during this session; therefore, installation commands, configuration fields, and differences between editors remain unverified, and secondary commands have not been reproduced. The coverage of marketing nodes is subject to the actual inclusion within the YiQiXiu Marketing Calendar; furthermore, a search URL is not equivalent to a template detail page. Data indicates zero installations, resulting in a lack of community experience for reference.
Read Moredsh-github-aide: Publishing Workspaces or Project Directories to GitHub within DSH
The `dsh-github-aide` plugin, maintained by maxid, integrates workspace synchronization and project upload capabilities. It features read-only pre-checks, secret scanning, generation of `.gitignore` files based on the tech stack, creation or reuse of GitHub repositories, git initialization and commits, as well as fetching, rebasing, and pushing to handle forks. Tokens are stored in the DSH credential store and utilized via an in-memory askpass process. The plugin supports `github_publish`, `github_status`, `github_configure`, and a sidebar Web panel. Requirements include DSH `>=0.2.1-rc.3 <0.3.0-0`, Node.js 22+, and git. Installation is available via `dsh plugin add link` or `github:maxid/dsh-github-aide#dsh-v0.2.1-rc.5`.
Read MoreSales Conversion Strategy Consultant: Codifying Sales Methodologies into Reusable Skills
Sales Conversion Strategist (business-ops skill @org-fq6oz6ky/sales-conversion) is a business-operations skill released by Chanmeng Technology that covers end-to-end sales-conversion solutions: product selling-point extraction, user persona definition, deal-close techniques, tiered customer follow-up, objection handling, and the creation of sales agents. Currently at version 1.0.0, with enterprise-class stability and certification, it has received an overwhelmingly positive response from 870 users, enjoys a flawless performance rating of 100000 review points, and has 0 production installs. The skill requires no API key for activation, aligning with a no-Code workflow (its invocation limit or whether it is commercial remains unclear and requires clarification). The article provides an in-depth overview of its positioning, core capabilities, and enablement path, and walks through real use cases such as sales-agent creation, objection handling with data verification, and shareable lead follow-up templates. However, enterprise security and compliance still need to be verified according to your deployment model before implementation.
Read MoreRapid Estimation Skill of Project Cost: Change the bid quotation calculation from temporary calculation to fixed procedure
gongcheng (Rapid Construction Cost Estimation, @user_fbd4533d/gongcheng) is a professional-class Skill published by the SkillHub community, version 1.0.0, requiring no API Key. It breaks down bid price estimation for county-level (Song County/Luoyang) housing construction, industrial plants, office buildings, glass curtain wall high-rises, rural revitalization, high-standard farmland, and comprehensive land consolidation projects into five steps: clarification of pricing assumptions, rapid estimation using the indicator method, detailed cost composition (work items/measures fees/statutory fees/taxes), sensitivity analysis of main materials, and output of bidding risk lists. This article introduces its positioning, capabilities and workflow, typical usage scenarios, and notes missing materials such as installation commands; it also indicates that the Vibecoding full-lifecycle Skill page fetch in the task clues failed and was therefore not adopted.
Read Moreai-engineering-legal: Integrating Engineering Cost and Legal Knowledge Bases into Agents
**ai-engineering-legal** is an AI Skill in the intersection of cost engineering and law, published by community author **user_23e3d4ab**. Based on the "AI Engineering Cost + Legal Knowledge Base" (1,394 subscribers, 3,149 items), it is categorized as **professional** (Version 1.0.0). Designed for four task categories—judicial appraisal of engineering costs, contract risk review, variation claim analysis, and settlement dispute diagnosis—it covers trigger keywords such as cost+legal, engineering judicial authentication, contract risk control, variation claims, settlement disputes, appraisal opinions, evidence chains, time extension claims, and cost claims. It also supports case analysis, simulated appraisal, and risk warnings based on the knowledge base. The skill is tagged with **requires_api_key=false**. This article details its positioning, core features, activation methods, and applicable limitations, while faithfully noting conflicts in ratings and download data across different sources.
Read Moredsh-ext-master: Manage Document Preview Extensions for DeepSeek Harness
dsh-ext-master is a DSH client plugin maintained by maxid. It uses the profile bundle and client mechanisms to append configurable extensions for document preview, treating `.mdc` as Markdown by default and supporting the extension of safe suffixes such as mdx in `cordis.patch.yml`. It is compatible with DSH >=0.2.1-rc.3 <0.3.0-0, and installation via `dsh plugin add` following the official tag is recommended. It does not modify `app.asar`; configuration takes effect through `__DSH_DOCUMENT_PREVIEW_EXTENSIONS__` injected by the Host. It is suitable for developers using `.mdc`/`.mdx` who require normal sidebar preview. Please review the source code and MIT license before installation.
Read Moredsh-config-studio: Form-based editing of non-volatile plugin configurations in Profiles
dsh-config-studio is a DSH client plugin maintained by MaxHan0102. It browses addressable Loader entries in the current profile via `ctx.configEditor`, renders forms based on each plugin's Config schema, and writes non-volatile fields to `cordis.patch.yml`, with the entry point located at Settings -> Plugin Configuration. The article explains its differences from `settings.mutate` (which only supports volatile fields), the browse/edit/save workflow, read-only rules, and the behavior of entire patch segment replacement. It also provides official installation commands and typical usage for both Desktop and dsh plugin. Requirements include DSH 0.2.1-rc.5+ and Node 20+, under the MIT License.
Read MoreProfessor Peng Kaiping - Positive Psychology: A Skill for Acting as a Thinking Advisor from a Positive Psychology Perspective
Professor Peng Kaiping – Positive Psychology: This is a thinking advisor Skill deeply distilled from Professor Peng Kaiping's public content, released by community developer user_d1e1e39e on SkillHub under the "knowledge-management" category (version 1.0.2). Covering eight books, ten Chinese and English papers, primary content from his Douyin/WeChat Official Account/podcasts, public course systems, and long-form video interviews, it employs a weighted hierarchy for first-hand and secondary sources, explicitly excluding Zhihu and encyclopedias. The Skill is triggered when users mention "Peng Kaiping's perspective" or positive psychology-related concepts, making it suitable for analyzing issues such as happiness, anxiety, involution/lying flat, education, and the relationship between AI and human nature. Please refer to the official page for installation instructions; note that secondary content requires cross-verification.
Read Moredsh-plugin-opencode-local: Leveraging All Local OpenCode CLI Models in DSH
Maintained by MateenAhmed737, this DSH model inference plugin integrates all logged-in providers and free models from OpenCode via a local `opencode serve` instance. Authentication and traffic are not proxied through the plugin. It is compatible with DSH 0.2.x and OpenCode 1.14.19+. To install, first install `opencode-ai` globally, then add `dsh-plugin-opencode-local` via the DSH plugin manager or CLI. It supports `serverUrl` attachment, DSH approval for restricted tools, streaming, and image input. This is suitable for developers who wish to reuse OpenCode locally, similar to T3 Code. Please review the source code and the MIT license before installation.
Read MoreData Management Expert: A Chinese Skill That Clarifies Data Management Capabilities
Targeting data engineering, BI, and middle-platform teams, this article introduces the “Data Management Expert” Skill on SkillHub. Published by user_ce145526 (version 1.0.0), it comes from the community, is categorized under data-analysis, and carries the tag requires_api_key=false. Focused on the full data lifecycle—collection, storage, processing, analysis, and utilization—it provides eight consistently framed descriptions: definition and scope, importance, practices, strategic significance, domains, the DCMM capability maturity model, challenges and advantages, and implementation. The article outlines its positioning, coverage, and typical usage patterns. It also clarifies limitations: this is a knowledge-description Skill that does not connect to metadata or repositories and cannot perform actual inventory or assessment; official pages and API scraping failed, and the installation command has not been verified, so it is not provided; operational metrics are snapshot-based—rating 100000, downloads 857, installs 0, with another snapshot showing rating 99641, downloads 812. It is suitable for quickly obtaining structured, consistent descriptions when drafting proposals, reports, or training materials; before external delivery, content should be reconciled with current standards.
Read Moreego-browser: Browser Automation Skill for Windows
ego-browser is a Windows-oriented browser automation skill built on Playwright and CDP Accessibility. It provides Space isolation, multi-frame support, and a learn system, allowing Agents to drive the browser via SKILL.md conventions instead of rewriting scripts each time. Coordinate: @user_310a65be/ego-browser-win; Version: 1.0.1; Source: community; Category: ai-agent; No API key required. The article describes its positioning, verified capabilities, applicable scenarios, and limitations, and honestly notes that because the official page could not be retrieved, the installation command and usage examples cannot be verified. It recommends visiting the official URL to confirm.
Read MoreBid Response Matrix: Converting Tender Requirements into a Checkable Response Checklist
"Bidding Response Matrix" is a Skill (Version 1.0.0, category: business-ops) published by SkillHub community author user_9e6539c8. It is designed to extract mandatory, prohibited, and recommended requirements—as well as qualification, delivery, technical, and safety criteria—from tender documents or procurement needs to generate a numbered response matrix. It offers an optional comparison against bid response documents to itemize coverage status (covered, partially covered, or not covered). The tool includes a local Python script that exports data as Markdown, JSON, or CSV. It does not rely on external APIs and requires no API key. It is suitable for bid preparation, bid inspection, qualification verification, and the identification of bid rejection risks.
Read Moredsh-cyber-dept: Turn Agent Teams into a Pixel Office on the Right Side of the Conversation
dsh-cyber-dept (Cyber IT) is a DSH community workflow plugin maintained by Master-chai. It displays an interactive pixel office on the right side of the current session: employee nameplates, a task board, and team communication linkage. It supports write operations such as assigning tasks, interrupting, and immediate execution, all of which require confirmation. Compatible with DSH >=0.2.1-rc.5 <0.3.0-0, it requires Node.js and npm for local builds, then installation via an absolute link: path. MIT-licensed open source; not an official product. Intended for visual collaboration in Agent Teams; review the source code and license before installation.
Read Moredsh-session-batch-manager：Batch Session Management in DSH Sidebar and Session Dashboard
The DSH Web plugin **dsh-session-batch-manager** (v0.1.1), maintained by mars401, provides sidebar batch multi-select, batch archive/restore/physical delete via a floating toolbar, and a full-screen Session management panel. It visualizes models, four-bucket tokens, cache read/write, and context pressure. The installation command is `dsh plugin --profile web add github:mars401/dsh-session-batch-manager`, requiring DSH >= 0.2.1. It is suitable for developers who need to batch clean sessions or get an overview of local session resource usage. Note that the plugin runs with the current dsh process permissions; please review the source code before installation.
Read More