dsh-codex-chat: Round Separators and User-Input Cards for DSH Web Sessions
dsh-codex-chat is a pure-style DSH Web plugin maintained by Song2012 that improves multi-turn conversation readability with turn separator lines and user input cards on the right, while AI responses remain left-aligned plain text. Restarting DSH Desktop is required after installation via a local link path. It relies on host DOM class name suffixes and theme tokens, is released under the MIT License, and has been tested on DSH Desktop 2.0.18.
Read MoreTencent Docs: Transforming Spoken Workflows into Flowcharts and Mind Maps in Enterprise WeChat
Introduce the Skill "Tencent Docs Voice-Generated Flowcharts and Mind Maps" (@hotelai/wecom-mindflow, v1.0.0, produced by Xie Aitong Studio, categorized as office-efficiency). Users verbally describe business processes or knowledge structures, and the AI organizes them into structured content. It first generates documents in WeCom smart documents, then syncs to Tencent Docs to create editable Chinese mind maps and flowcharts, and finally embeds canvas links back into the WeCom document for team collaboration. The article covers core features, coordinate-based activation methods, typical usage workflows, and notes such as the requirement that both WeCom and Tencent Docs must be available simultaneously.
Read Moredsh-sap-ai-core: Integrating SAP AI Core and cn40 Qwen Multi-Model in DSH
dsh-sap-ai-core is a DSH model inference plugin maintained by solstice621 that integrates with SAP AI Core Orchestration under a shared OAuth premise. Version 0.1.0-rc.4 supports five Qwen models for cn40, five thinking budget levels, and OpenAI-compatible Chat Completions; the plugin handles token management, deployment routing, streaming, and tool protocols. The article details installation commands, cordis.patch, cn40 preset configurations, diagnostic CLIs, as well as applicable scenarios and permission considerations.
Read MoreResume Inspector — 18-Dimension Matrix Scoring & Item-by-Item Rewriting: First, score with scripts; Then, let LLM rewrite only the phrasing
Resume Scorer (@user_04c33885/resume-scorer) is a SkillHub community skill (Version 1.0.0) categorized under content-creation that requires no API key. Built upon authoritative job-seeking guidelines from institutions like the Harvard Career Center, it establishes a scoring matrix of 18 points derived from 6 major modules and 3 quality dimensions, assigning scores item-by-item for a total of 100 points. The architecture employs a hybrid approach of deterministic script scoring combined with LLM-focused suggestions: structure, quantification, and verb usage are determined by scripts with zero token consumption, while the LLM provides supplementary comments on the expression level. For modules with low scores, it directly offers side-by-side comparisons of original and rewritten sentences utilizing strong verbs and data points, rather than offering vague advice. It supports PDF, Word, and plain text formats, automatically recognizing complex layouts such as two-column designs and triggering a full evaluation. Trigger scenarios include requests to "score my resume," "review my resume," "optimize my resume," as well as "resume review," "rate resume," or "evaluate resume," or when a user directly provides a resume file for assessment. This skill is suitable for self-assessment by individual job seekers and for embedding resume scoring capabilities into products. Note that image input is not listed in the official documentation. As this is a relatively new skill, it is recommended to validate its effectiveness with a sample resume before extensive use.
Read MoreBan Xian’s Eight-Character Reading: Delegating Eight-Character Chart Casting, Pattern Analysis, and Climate Adjustment Analysis to Skill
The "Immortal Bazi Fortune Teller" is a community Skill published on SkillHub (@user_2c1658eb/bazi-fortune-telling, v1.0.1). Based on classical texts such as *Ditian Sui*, *Ziping Zhenquan*, *Qiongtong Baojian*, *Qianli Minggao*, and *Shenfeng Tongkao*, it provides Bazi chart casting, Day Master analysis, Pattern analysis, Climate adjustment analysis, and brief analysis of Major Luck and Annual cycles. The page displays a rating of 100,000 and 857 downloads, with no API key required. The article introduces its positioning, core features, activation methods, typical usage, and applicable precautions. It emphasizes that this tool is for metaphysical deduction rather than statistical prediction, and users should verify it independently before integration.
Read Moredsh-opencode-go-auth: Integrating OpenCode Go Subscription into DeepSeek Harness Desktop
The community plugin dsh-opencode-go-auth provides a native settings page for OpenCode Go / Go Plus in DeepSeek Harness Desktop: it validates and stores the API key in the credential store, displays 5-hour / weekly / monthly quotas, refreshes usage and the model directory in the background, and lets you check conversation models under Model Display. Install address: github:solstice621/dsh-opencode-go-auth#main；in-session connection name: OpenCode · Go quota. MIT License, primary testing environment: macOS. The plugin does not increase quota and has no affiliation with OpenCode official.
Read MoreIndustry-Specific Tax Expert Panel: Delegating Sector-Specific Tax Matters to Dedicated Specialists
**# Preface General tax Q&A models often lack industry depth: when discussing “input VAT deduction,” metallurgical enterprises and pet retail chains have completely different concerns; when discussing “deductible cost recognition,” traditional Chinese medicine clinics and construction contractors also require very different answers. If you want judgments that closely align with industry practice from a single query—rather than getting a lengthy background explanation before a generic conclusion—the Skill introduced below is worth checking out first. # What Is This? In one sentence: This is a tax-focused Skill organized as an “Industry Tax Director + several industry-specific subject experts,” designed to cover deep tax issues in vertical industries. - Name: Industry-Specific Tax Expert Panel - Location: @user_11064e10/tax-expert-industry-specialty - Version:**
Read Moredsh-conversation-cost: Estimating per-turn conversation API costs in the DeepSeek Harness Desktop
<trans><fragment>The community plugin dsh-conversation-cost automatically syncs model per-unit prices for DeepSeek Harness desktop, displays a Cost figure between usage and time for each response, and shows the cumulative estimated USD expense of the current conversation below the input box.</fragment><fragment>It supports token breakdowns across four categories, multi-model price tiers, and an offline price cache, with no additional API key required.</fragment><fragment>Install at github:solstice621/dsh-conversation-cost#main; amounts are for cost-scale reference only and do not constitute platform billing.</fragment></trans>
Read MoreAutobiographical Archivist: Organizing Fragmented Life Materials into a Traceable Autobiography
"Autobiography Reconstructor" is an organizing skill released by the SkillHub community, listed under the knowledge-management category (located at @user_00c9b356/autobiographyreconstructor, v2.1.0). It faithfully organizes source materials—including diaries, notes, letters, oral histories, and photo captions—into autobiographies or memoirs with traceable destinations. Its core principle is to faithfully organize without fabrication: every piece of final text must feature clear source tags. Standout elements are revealed by supplementing verifiable historical backgrounds (such as reframing perspectives), rather than psychological analyses or fictionalized scene descriptions. The workflow consists of: extracting source materials paragraph by paragraph → assigning source tags → extracting standout elements via perspective shifts → structuring content into chapters → embedding corresponding images → exporting in multiple formats. The output includes source citations, image embeddings, and an anti-fabrication checklist. It exports in Markdown, DOCX, and PDF formats, and requires no API keys. It minimizes friction for organizing oral histories, diaries, and family histories, but is not designed for pure fiction, simple format conversions, or modifications of pure text. The documentation transparently notes that the official page is currently inaccessible and the provided materials lack installation commands.
Read MoreLanding Optimizer: Directing Influencer Traffic to Conversion-Optimized Pages
Landing Optimizer is a business-ops Skill published by the SkillHub community, currently at version 19.0.0, and requires no API Key. Designed for optimizing landing pages prior to the launch of influencer campaigns, it delivers an information matching audit, recommendations on page structure and trust elements, conversion strategies for promo codes and CTAs, as well as an A/B testing roadmap; it does not handle post-launch reviews (handled by performance-analyzer). Users trigger it by describing their goals in natural language.
Read MoreDSH Better Answer View: Embedding Charts, Cards, and Interactive Responses in Conversations
DSH Better Answer View is a DSH client plugin maintained by solstice621 that embeds charts, cards, and interactive blocks into the chat body using tools like `render_answer`. It stores HTML and state in a local `web-answer` directory and supports revision history. The plugin can be installed via `github:solstice621/dsh-better-answer-view#main` and has been validated on Harness Web 0.1.5-rc.3 and Desktop 0.2.1-rc.3. It is well-suited for comparisons, trends, and lightweight parameter tuning, while simple Q&A remains in plain text.
Read Moredsh-qa-jump: Jump to Question/Answer and Code Block Folding in DSH Dialogs
dsh-qa-jump is a DSH client plugin maintained by sol-ray that adds dual-axis Q&A navigation and jump-to-previous/next entry features above the Web GUI input box. The jump target aligns with the start of the answer (including the completion time row). During streaming generation, it restores scroll control via synchronous positioning and the scroll event. It also provides a pure CSS one-click collapse for fenced code blocks. Based on the GitHub README and package.json, this article provides npm/GitHub installation commands, a usage table, and an explanation of DOM hook dependencies, advising users to review the source code and process permissions before installation.
Read MoreMedical Penguin - Medical Communication Support: Transforming Package Inserts and Literature into Reusable Medical Q&A Assets
"Pharmaceutical Penguin - Medical Communication Support" (Username: @org-gfymbxku/wmqe-pharma-ma-comms, version 1.0.4, Professional category, published by Weiming Penguin (Beijing) Technology Co., Ltd.) on SkillHub is designed for pharmaceutical industry Agent developers. It constructs a medical FAQ library based on drug package inserts and literature, providing three core capabilities: clinical needs analysis and English medical literature translation. The skill is tagged with requires_api_key=false, indicating no additional API key configuration is needed. The article explains its positioning, three core functions, and enablement constraints based on the universal SKILL.md format. It also points out that information regarding official installation commands and platform-specific differences is currently missing from available materials and requires independent verification. Furthermore, it cautions that medical content released externally must undergo professional review.
Read Moredsh-codex-usage: View Codex quota in DSH sidebar and bottom bar
dsh-codex-usage is a DSH networking tool plugin maintained by sodaa-l. It displays the remaining 5-hour and weekly quotas, reset times, and refresh policies for ChatGPT Codex in the sidebar and bottom bar. It logs in via OpenAI OAuth, and the host requests the wham/usage API endpoint. The article explains the Web/Desktop installation commands, the toggle switches in the settings page, and how to use the bottom bar pop-up. It also reminds users of requirements such as the non-public API, process permissions, and DSH >=0.2.0-rc.1.
Read MoreIntroduction to the Customer Profiling Capability: Transforming Imprecise Customer Data into Actionable, Dynamic Profiles
Released by Kunlun Growth, the Customer Profile Skill (repo: @kunlungrowth/kunlun-customer-profile, version 1.1.0) is categorized under business-ops and requires no API key. It transforms raw customer data into actionable profiles, covering the entire pipeline: data collection → tag taxonomy → segmentation → scenario matching → operational actions. Built upon three core methodologies—the RFM model, tag tree, and segmented operations—it comprises five capability modules. Official documentation indicates that outputs are ready for direct delivery, and the workflow includes troubleshooting guides for bottlenecks alongside compliance guardrails. The skill is applicable to concepts such as user personas, customer segmentation, audience profiling, and buyer personas. The provided materials do not include official installation commands or reproducible examples; the text acknowledges this transparently and directs users to the official repository.
Read MoreAcademic Paper Self-check: Multi-dimensional Self-check Before EAS Course Paper Submission
Academic Paper Self-check is a community Skill (version 1.0.0, categorized under education) published by user_a722fc01 on SkillHub, designed for pre-submission self-review of the semester's Research Paper (Literature Review) for the EAS course. Derived from Course Unit 14 (From Draft to Manuscript), it systematically checks and provides revision suggestions across eight dimensions: abstract quality, title refinement, vertical alignment, terminology consistency, evidence integration, tense and modal verbs, APA citation accuracy, and figure and table data formatting. No API Key is required. The Skill is activated by trigger phrases such as "check my paper" and "review my research paper". This article outlines its features, activation methods, typical use cases, and applicable limitations.
Read Moredsh-folder-launcher: Add Folders to DSH Web Workspace from File Explorer
dsh-folder-launcher is a DSH client plugin maintained by snnnke that registers externally provided folder paths to the running DSH Web sidebar via workspaceRegistry. On the Windows side, it provides an Explorer right-click "Add to DSH Web" without administrator privileges, calling POST /dsh-folder-launcher/add via loopback and bearer token. If DSH Web is not running, it can automatically start it. Linux/macOS users can use bin/dsh-folder-launcher.mjs. The installation command is `dsh plugin --profile web add github:snnnnke/dsh-folder-launcher#v0.1.0`, and the context menu requires separately running install-context-menu.ps1 in the repository root directory.
Read MoreAipDOOH: Consolidating pDOOH Operations into 24 MCP Tools
AipDOOH is an Agent Skill designed for the pDOOH business, exposing 24 tools via MCP Proxy. It covers outdoor advertising location queries, campaign plan generation, ROI calculation, competitor analysis, and market intelligence collection, supporting direct invocation via Python and Node.js. Published by the SkillHub community account user_437f7810, it originates from the 'community' source and is categorized as 'business-ops', with version 1.0.1 and a page tag of requires_api_key=false. The server address must be injected via the environment variable PDOOH_HOST; the Skill does not contain any real server addresses, requiring users to self-deploy or integrate with their own backend. This article describes the capability layering, access methods, and applicable scenarios. It also transparently marks the installation commands and sample codes as unverified due to a failure in scraping the official page, recommending a direct review of the official page for confirmation before integration.
Read MoreAI Stand-up Voice: Authentic Opinions without Pretense — Removing the AI Flair from Chinese Responses
This is a Chinese content creation-oriented roast expression Skill, located at @user_f7d35355/ai-standup-roast-root, version 1.0.0, community-sourced. Its publisher is user_f7d35355, categorized under content creation, with the tag *requires_api_key* = false. Instead of simply replacing words, it adjusts the tone of expression: it transforms ordinary responses into a sharp-tongued yet polite roasting style that speaks boldly with clear boundaries, avoids replicating performer stunts, and refrains from quoting scripted lines. Furthermore, it supports original comedic twists and callbacks within the given context. The article introduces its positioning, four core constraints, the general SKILL.md loading method (since the official team has not provided a dedicated installation command, only the general path and tag meanings are included), the input format for its rewriting-based usage approach, applicable scenarios (such as social media copy, comment replies, and rewrites for workplace and product contexts), as well as key notes: it only modifies the tone rather than altering facts, callbacks strictly depend on the context, and boundaries are left to the model's own discretion to manage. When it was retrieved, the page displayed a rating of 100,000, 874 downloads, and 0 installations; these figures should be subject to the actual page at the time of access.
Read MoreRenovation Consultant: Line-by-Line Auditor for Quotes and Contracts
Renovation Advisor Pro is a general-purpose Skill in SKILL.md format released by the DouEase team, available at @douease/renovation-advisor-pro (v1.0.0, knowledge-management). It serves as a private renovation advisor and an expert for reviewing quotations and contracts: given a quote or contract text, it returns item-by-item risk analysis, an overall estimate of price padding, and negotiation recommendations. The skill covers key process stages including quotation review, scanning for onerous clauses in renovation contracts, electrical and plumbing works, waterproofing and water-retention testing, masonry and tile laying, carpentry and ceiling installation, painting, and completion acceptance. It can identify upsell add-ons, missing items, underquote-then-hike pricing, unit traps, brand swapping, absence of delay-compensation clauses, and unbalanced payment schedules. No API key is required; it is ready to use right after installation and is driven purely by methodology. This article introduces its core features, official location, and usage.
Read MoreAgentizer-zh: A Reverse Trace Removal Skill for Highly AI-Flavored Content
Agentizer-zh is the satirical Skill that serves as the mirror counterpart to Humanizer-zh. Rather than stripping away AI tone, it injects rhetorical negation, three-part structure, tech metaphors, abstract philosophy, and open-ended conclusions, rewriting any text or dialogue into heavily AI-flavored prose for comedic and critical effect. Published by user_c1337c02 under the community source in SkillHub's content-creation category, the current version is 1.0.1, located at @user_c1337c02/agentizer-zh, tagged as requires_api_key=false, meaning no API Key is needed for invocation. The page records a rating of 100000 and 862 downloads. Two entry points are provided: /agentizer-on and /agentizer-off to toggle a continuous dialogue mode, and /agentizer [text] for single-instance rewriting, with command names subject to the official page documentation. The article outlines its positioning, verified capabilities, activation method, reproducible usage examples, and applicable scenarios and limitations—output is satirical and unsuitable for formal documents, and continuous mode will override subsequent output, requiring manual shutdown after use. Official URL: https://api.skillhub.cn/user_c1337c02/agentizer-zh.
Read MoreDSH-Reason-Tree: Decomposing Conclusions into Argument Trees and Tracing Back to Source Text in DSH
dsh-reason-tree (root cause tracing) is a DSH model reasoning plugin maintained by Snerou. It represents "conclusion←reason" trees using Markdown indented lists, where nodes can be marked with [+]/[-]/[?] and written back to files. Line ends can associate memories or file paths, and when combined with dsh-hippocampus, it can "locate to hippocampus." The default directory is the workspace trace, and dir and readRoots can be configured via ~/.dsh/dsh-reason-tree.json. Install with: dsh plugin --profile desktop add dsh-reason-tree. MIT licensed, with data remaining on the local machine.
Read Moredsh-hippo: Visualizing Workspace, Memory, and Conversation as Rotatable 3D Neuronal Graphs
dsh-hippo (hippocampus) is a DSH memory plugin maintained by Snerou that read-only scans workspace Markdown, ~/.dsh memories, and session/trace md files, then renders, in the sidebar, a scalable and rotatable neural-network visualization with files as clusters, headings and list items as nodes, and keywords as edges. It supports installation via the desktop profile, optional dsh-hippo.json for additional data sources and session limits, local zstd caching of sessions, and integrated locating with dsh-reason-tree; all data stays local, and it is released under the MIT license.
Read Moredsh-genui-scope: Hierarchical Trimming of GenUI Card System Prompts
dsh-genui-scope is a DSH plugin maintained by SnakeDream that, when dsh-genui is enabled, trims GenUI component descriptions at the system-prompt/assemble stage at three levels (global, workspace, and session), reducing a fixed overhead of roughly tens of thousands of characters per turn and lowering component misselection. It supports presets such as lean/text/form and the card_components tool, and requires Harness 0.2.1-rc.5 with a local directory installation followed by a restart. It is open source under MIT.
Read MoreTheme Class Meeting Design: Turning Coherent Descriptions into Actionable Frameworks Designing Theme Class Meeting Plans: From Coherent Text to Actionable Implementation Structures
Targeting class meeting scenarios, this Skill is released by Kunlun Growth (located at @kunlungrowth/kunlun-edu-class-meeting), categorized under education with an enterprise source (Version 1.1.0). It limits class meeting outputs to four sections: objectives, processes, activities, and courseware key points, covering themes such as moral education, safety, and psychology, with a specification of requires_api_key=false. The article outlines its positioning, verified capabilities, the necessity to confirm loading methods based on the specific Agent used prior to activation, and a usage scenario based on public introductions. It also clarifies that preliminary proposals cannot replace teacher judgment, and content regarding safety and psychological themes requires review. Data on the official page is subject to dynamic changes.
Read More