transparent-encrypt-file-mcp: Enabling Agents to Read and Write Plaintext Files in a Transparent Encryption Environment
On Windows desktops deployed with transparent-encryption DLP solutions such as ESAFENET, Tianrui Lingdun, IP-Guard, and DSE, the AI Agent's built-in Read/Write functions can only obtain ciphertext. This community Skill provides a Node.js MCP Server (40 tools) that delegates read/write operations to DLP whitelisted processes (Office/WPS/Outlook/AutoCAD COM, 7-Zip, LibreOffice), returning plaintext while keeping write-out controlled. It includes a capability_matrix for capability queries, a doctor self-check, diagnostic workflows, and byte-level forensics via forensics_headers/inventory_encrypted/compare_plaintext/audit_trail. The pywin32 channel in scripts/*.py covers PPT speaker notes, Access rows, WPS cells, Acrobat editing, and Office-to-PDF conversion. It is only suitable for Windows interactive desktops, and is not suitable for environments without DLP, scanned PDFs, or files that are already readable.
Read MoreOperations: Turning Review, Menu, and Local Search Data into Actionable Recommendations
# Preface Catering teams eventually encounter problems like these: reviews from Meituan, Dianping (大众点评), and Xiaohongshu (小红书) are scattered across platforms, leaving teams unsure what customers are really complaining about; menus expand endlessly, yet gross profit margins remain unclear; stores fail to rank well in local searches, and it is hard to know which specific aspect to change. Manually reviewing each evaluation and manually calculating the menu structure is labor-intensive and costly, and the resulting conclusions are often inconsistent. Introduced below is an Agent Skill targeted at such scenarios, named 'Operations' (运营), developed by community developers and published on SkillHub. It delegates three types of tasks—review analysis, menu optimization, and store diagnostics—to a model, producing actionable operational recommendations. # What is this Positioned in one sentence: an operational analysis Skill targeted at catering stores, covering three directions: review analysis, menu optimization, and store diagnostics. Development and Maintenance: The publisher...
Read MoreJob Competency Modeling Assistant: Generating Unified Competency Standards for a Position
Introducing the SkillHub community Skill “Job Competency Modeling Assistant” (@user_fa14b5bc/competency-model-builder, version 1.0.0, category business-ops). This Skill uses a five-layer breakdown of the Iceberg Model combined with 5-level behavioral anchoring to produce a complete competency model for a specified position, including an interview question bank, development pathways, and implementation guides for four scenarios, providing unified standards for hiring and talent development. The tag is requires_api_key=false; no API key is required. The article explains the core features, how to enable it based on SKILL.md (no specific installation command has been provided by the official source, and no speculation will be made), typical usage, applicable scenarios, and cautions, and provides the official URL.
Read MoreDrug Sales: In-hospital Prescription Drug Sales Execution
drug-sales is a community Skill designed for the execution of in-hospital prescription drug sales scenarios (@user_31815b66/drug-sales, v1.0.0, Professional category). It integrates seven core capabilities into a single execution pipeline: customer and terminal management, regional planning and KPI operations, professional visits and academic promotion, pure sales and commercial channels, POA management, compliance red lines, and data-driven SFE. It also features a built-in foundational module (foundation.md) that reuses data calibration rules and clinical profile frameworks. Boundary demarcation is a key design element: this Skill acts as the business execution party, while drug-training serves as the training content provider, and drug-hr handles staffing and productivity. Compliance red lines are managed by compliance teams, and medical evidence by post-marketing medical teams. The interface depth adapts to user proficiency, switching between a novice mode (detailed guidance + input checklists) and an expert mode (concise output + direct conclusions). Typical triggers include phrases such as "How to break down KPIs for this region?", "How to investigate a drop in pure sales?", "How to schedule department meetings?", or "What are the compliance red lines for representative behavior?" The Skill does not require an API Key to run (requires_api_key=false) and utilizes Alipay for distribution (pricing_type=alipay). As official documentation does not provide executable installation commands, loading steps for specific clients should follow the official address page instructions.
Read Moredsh-scoped-http-proxy: Explicitly Invoking HTTP Proxy Service
dsh-scoped-http-proxy provides DeepSeek Harness with the optional Cordis service `scopedHttpProxy` and a Web settings page. It utilizes the configured HTTP(S) proxy only when plugins actively invoke `fetch`, without altering system proxies, environment variables, or the global `fetch`. This article introduces installation commands, `proxyUrl` configuration, the `inject` integration method, as well as collaboration boundaries and security considerations regarding Fake-IP plugins.
Read Moredsh-session-bar: Making DeepSeek Harness Sidebar Sessions Speak for Themselves
dsh-session-bar is a Web client plugin for DeepSeek Harness that displays estimated completion percentage and a 6px progress bar in each sidebar row for running tasks. Upon completion, the entire row turns orange and remains until clicked. In case of timeout, it switches to displaying the actual elapsed time. The installation command is `dsh plugin --profile web add dsh-session-bar`. It requires zero configuration, with state persisted in localStorage. Compatible with DSH 0.2.x; coexists with dsh-session-keeper row structure; MIT licensed.
Read MoreDecoration Quantity Surveying · Budget Estimate & Material Planning: Assigning Decoration Estimation and Material Planning to Agents
"Decoration Quantity Takeoff · Budget & Material Planning" is an Agent Skill from the SkillHub community (Coordinate: @user_43faedd6/zhuangshi-suanliang, Version: 0.9.31, Source: community, Category: professional, Rating: 100000, Downloads: 1296, requires_api_key=false). It rapidly generates decoration budgetary estimates and material plans based on room types, architectural descriptions, architectural drawings, structural drawings, room decoration schedules, construction method tables, and node detail drawings. It can import length and area measurement candidates from the cad-file-reader for verification, but construction method mapping, opening deductions, material loss, material ledgers, and bill quantities are calculated by the skill itself. This article introduces its input/output boundaries, division of labor with cad-file-reader, activation methods (since the official page does not provide installation commands, metadata is the standard), applicable scenarios, and precautions.
Read Moredsh-opencode-usage: OpenCode Go Per-Model Usage & Multi-Account Settings Page
The DSH client plugin dsh-opencode-usage (dsh-opencode-usage-settings 1.4.1) displays OpenCode Go usage and remaining quota by model for the current month/week on the settings page, reads the dsh-opencode-quota-router account pool in read-only mode, and manages multiple accounts on the same page. Usage is self-calculated from the last 40 days of sessions logs and session/event data. The Host provides GET /opencode-usage.json. The README does not provide installation commands; note that it only counts this harness, amounts are estimates, and the interface has no authentication. MIT licensed, maintained by 2869949486.
Read MoreScientific Thought Shrimp: Turning the Five-Step Closed Loop of Experimental Science into a Reusable Skill
# Introduction Researchers often encounter specific challenges: discovering uncontrolled variables halfway through designing an experiment, realizing the initial hypothesis was unclear only after data analysis is complete, or struggling to find a stable, executable process for training scientific thinking. General dialogue models tend to provide direct answers to such problems rather than guiding users through the process step-by-step. "Scientific Thought Shrimp" (@user_a0501ee4/scitf) is a community Skill designed to address this pain point. It solidifies a fixed action framework, enabling the model to guide users through a complete thinking and experimental workflow whenever necessary. # What is this Scientific Thought Shrimp is a Skill for scientific thinking training and experimental research guidance based on the "Pentacyclic Action Framework of Experimental Science." It was released by community developer user_a0501ee4, and the current version...
Read MoreSeedance Video Generation By tokease: Let Agents Generate Video from Text
This article introduces the community Skill "Seedance Video Generation By tokease". The skill is located at @user_92ca4468/seedance-video-generation-by-tokease, published by user_92ca4468, categorized under design-media, with a document version of 1.0.0. It is positioned to handle requests such as "generate/create/produce videos, AI videos, text-to-video", and accomplishes text-to-video generation by invoking the Seedance 2.0 model through the tokease.com API. The scraped data shows the tag requires_api_key=true, so a valid tokease.com API Key needs to be prepared and properly managed before use. This article explains its functional boundaries, matters to confirm before enabling, and typical natural language usage; the article does not provide verifiable installation commands, configuration fields, or output parameters, and related details should be based on the official page.
Read MoreSocial Security and Human Capital Tax Expert Group: Deconstructing Social Security and Tax Compliance into a Multi-Role Collaborative Skill
Community Skill, coordinated by the Director of Social Security and Human Capital, brings together specialists in personal social security planning, social insurance tax inclusion, and declaration compliance. It focuses on compliant management of enterprise labor costs and risk management related to social insurance and taxation. Located at @user_11064e10/tax-expert-social-insurance-hr, Version 3.18.64, categorized under professional, with a rating of 100,000, 1,112 downloads, 0 installations, and no API key required. The article explains its capability structure consisting of the coordination layer, topic layer, and objective layer, and notes that the official documentation does not provide installation commands or usage examples. Before integration, users are advised to conduct their own testing and verify tax-related treatment.
Read MoreAmazon Consumer Insights · In-depth Demand Report: Transforming Reviews into Traceable Demand Conclusions
This article introduces a community-sourced "Amazon Consumer Insights · Deep Demand Report" Skill (published by user_e02a09ad, version 1.4.12, categorized under data-analysis). The Skill comprises two main capabilities: First, generating in-depth consumer insight reports covering user personas, purchase drivers, usage scenarios, pain points and unmet needs, competitive substitution relationships, and improvement opportunities, with a requirement to trace every conclusion back to specific review evidence. Second, providing comprehensive ARI review analysis capabilities, including subscribing to ASINs for review collection, browsing and filtering reviews, viewing star ratings/keywords/trends/radar charts, generating VOC and deep insight reports, product operation workflows, product change monitoring, negative review alerts and a negative review workbench, industry benchmarking and category ranking, as well as exporting reviews to CSV and reports to Markdown/HTML. It requires an ARI API Key (ari_live_*). The article explains installation and activation methods, typical reproducible use cases, and applicable scenarios, while noting that installation commands are not provided in the materials and that community validation is limited due to low download volumes.
Read Moredsh-opencode-routing: Unified OpenCode Account Pool, Quota Routing, and Usage Dashboard
dsh-opencode-routing is a DSH model inference plugin that integrates the original quota-router and usage-settings features. It supports multi-account OpenCode pooling, automatic A→B failover and retry on quota or authentication failures, single-model isolation, and whole-pool fallback. The "OpenCode Accounts and Usage" section on the Settings page displays official account usage percentages and local estimates per model. Install via `npm i dsh-opencode-routing`; it includes `cordis.patch.yml` and provides `/opencode-usage.json` and `/opencode-quota-router.json`. Please review the source code and the MIT license before use, as the plugin runs with dsh process privileges.
Read MoreAI Assistant for Registered HVAC Engineers: A Multi-Role Collaborative Assistant for HVAC Exam Preparation and Practice
This article introduces the Registered HVAC Engineer AI Assistant on SkillHub (@hongnan/hengwen-hvac-team, v1.0.1, enterprise source, published by Guizhou Hongnan Network Technology Co., Ltd.). Targeting exam preparation and professional work scenarios for HVAC engineers, the core mechanism of this Skill is switching between five expert personas based on instructions to collaboratively complete tasks. Classified as 'professional', it does not require a separate API Key for use. The article provides examples of usable prompts, while noting that details such as the division of labor among personas, installation commands, and activation methods for various editors are undisclosed. It reminds users to verify professional conclusions, warns against using the tool as a substitute for a registered engineer's signature, and explains the snapshot discrepancies in page ratings and download counts.
Read Moredsh-opencode-quota-router: Multi-account OpenCode Go Quota Routing and Model-based Fallback
dsh-opencode-quota-router is a DSH model inference plugin that automatically switches between multiple OpenCode Go accounts based on 5-hour, weekly, and monthly usage windows. When a specific model quota is exhausted, only that model falls back to the DeepSeek account, while other models continue to utilize the subscription. It mirrors the OPENCODE_API_KEY via an agent/request waterfall and restores the primary route through usage detection. Accounts can be managed via dsh-opencode-usage-settings or by accessing GET/POST opencode-quota-router.json. Installation: dsh plugin --profile <profile> add dsh-opencode-quota-router. MIT License, Maintainer: 2869949486.
Read MoreIntroduction to the AI Comic Skill: Breaking down novels, stories, or one-sentence ideas into storyboard scripts for AI image generation.
AI Manju (AI Comic Drama) (Coordinate: @user_c8b52036/aimage, version 1.0.0, released by the community, category: content creation, no API Key required) is a Skill based on the universal SKILL.md format. It converts novels, stories, or one-sentence ideas into storyboard scripts for AI image generation (AI comic dramas or vertical short dramas); additionally, it provides five-dimensional diagnostic scoring, character card packs, world Bibles, competitive benchmarking, and differentiation positioning. It is suitable for creating AI comic drama scripts, breaking down storyboards, assessing script viability, generating character setting cards, building world-building systems, or analyzing similar competitors and market positioning. The scraped materials do not provide installation commands or configuration fields; please refer to the official page for activation details. The article is organized into sections including the preface, what it is, core features, installation and activation, typical use cases, applicable scenarios and cautions, and the conclusion, without fabricating any cases or data.
Read Moredsh-reasoning-translate-zh: Translating reasoning into Chinese at the display layer
**Input Information** `dsh-reasoning-translate-zh` (package name `dsh-think-zh`) is a DSH client plugin that translates model reasoning into Chinese in real-time without altering the session or export settings. It displays a summary when collapsed and the full text when expanded, with an option to view the original English text. It does not inject system prompts. The engine supports DSH-configured models, Ollama, or OpenAI-compatible endpoints; the local `qwen2.5:1.5b` model is recommended. The installation command is `dsh plugin --profile desktop add git+https://github.com/2503455457-max/dsh-reasoning-translate-zh.git`. It is compatible with DSH >=0.2.1-rc.2 <0.3.0.
Read MoreProposition Tool: Automatic Question Generation Skill for Strongly Constrained Certification Question Banks Based on Textbook Slicing
Exam Item Generator (@user_2b70de78/exam-item-generator, v1.1.0, community, education) is a textbook-slice-based Skill for automatic exam item generation for certified question banks. It generates single-choice, multiple-choice, and true/false questions section by section based on textbook slices, employs strict out-of-scope hard-blocking mechanisms to prevent out-of-syllabus questions, supports tiered difficulty item generation and checkpoint resumption, allows results to be exported to Excel files with each question traceable to its textbook source. This tool is suitable for scenarios such as 1+X certificates, the Soft Examination, corporate internal training, and school question banks where textbook-based, high-volume exam generation with strict quality control is required. The author has used it to generate over 1,700 questions. Note: The rating and download count snapshots in the scraped data are inconsistent, and the installation command could not be verified online. Please refer to the official page for the most accurate information.
Read Moredsh-bilibili-retrieval: Integrating Bilibili Videos into the DeepSeek Harness Retrieval Layer
dsh-bilibili-retrieval is a DSH networked tool plugin released by maintainer 2332239652 (MIT license, v0.2.0, zero runtime dependencies). It provides nine tools including bili_search, bili_video, bili_comments, bili_frames, and bili_transcript, supporting search noise filtering, full description text retrieval, uploader pinned comments, segmented subtitle time anchors with preview sprite sheets; subtitle and AI summary features require QR code login. To install locally, use `dsh plugin --profile desktop add link:path`, then restart and verify with bili_probe. Credentials are stored in DSH_HOME, and the plugin operates in read-only mode for accounts.
Read MorePharmacy Drug Information Push Assistant: Converting Drug Data into Sales Scripts
This article introduces the community Skill "Pharmacy Product Promotion Assistant." By inputting a drug name and optional selling points, the AI queries drug information online, generates customer-facing sales scripts, and outputs PPT-style images for WeChat push notifications upon confirmation. The article explains the tool's positioning, core functions, trigger methods, usage, and compliance considerations, and provides the official address.
Read MoreStore Renovation Renderings with Nano Banana Pro: Converting Store Photos into Customer Flow Layouts
This article introduces the SkillHub community Skill "Nano Banana Pro Store Renovation Rendering Generator." Based on the SKILL.md format, this Skill is produced by user_113322ce, categorized under design-media, and is tagged with requires_api_key=false (Version 1.0.2). The Skill accepts text descriptions of business type, brand style, and functional requirements, and optionally accepts store photos or floor plans as references. It utilizes Nano Banana Pro to generate renovation renderings that showcase store traffic flow and brand experience, outputting validated local images. It is suitable for requests such as "Use Nano Banana Pro to create store renovation renderings for me"; however, if designing only the window area, the "Window Display Rendering Generator" should be used instead. This article outlines core features, organization of typical inputs, applicable scenarios, and limitations. It also notes that if installation commands are not provided in the documentation, the SkillHub page should be treated as the standard.
Read MoreLawyer Workbench Expert: Aggregating Cases, Schedules, and Deadlines into a Local Three-Column Page
The Lawyer Workspace Expert (@user_b8af0005/lawyer-workspace-expert, v1.1.0, community-sourced, professional category, requires_api_key=false) is a Skill that generates a personal workspace where "all files stay local": it consolidates case folders, Excel schedules, litigation deadlines, and work notes into a three-pane page you can open and view instantly, while files remain where they are and are not uploaded to the cloud. It covers modules such as case-folder hooking, read-only Excel integration, litigation-deadline integration, a case dock, calendar sync, itinerary capture, and desktop icons. It can be triggered in natural language, e.g., "I want to create a workspace," "Put cases and schedules together for viewing," "Case folders are too deep; I want one click to open them," "Litigation deadlines are easy to miss," or "I manage cases in Excel and want to turn it into a kanban board." It can also add modules to an already-built workspace. Official materials do not provide a universal installation command or IDE-specific differences; refer to the official address page for details.
Read Moredsh-meoo-connect: Integrating Meoo Models into DSH Sessions and Settings
dsh-meoo-connect is a community MIT-licensed plugin that automatically syncs the model catalog from the Meoo project service AK via OAuth, writes the providers.meoo route into llm-pi-ai, and makes Meoo Miaowu appear in the dsh session model selector and Settings → Models. It does not require a local proxy; streaming and tool calling use the native dsh pipeline. To install, run the official add command from a web profile, add dsh-meoo-connect to bundles, and restart; usage centers on /meoo login and /meoo sync. Intended for developers who already have a Meoo project and cloud services; review the source code and process permissions before installing.
Read MorePaper Abstract Extraction: Compressing Long Papers into Structured Key Points
Paper Summary Extractor (user_138a3c0c/paper-summary-extractor) is a community Skill published by user_138a3c0c on SkillHub, categorized under knowledge-management, version 1.0.0. It accepts paper links, paper text, or files, automatically retrieves the content, and extracts the title, authors, research questions, methods, experimental results, and innovation points in a hierarchical manner, outputting the information in a structured format. To use it in OpenClaw, send a paper PDF link or text and say "Summarize paper," and the Skill will streamline the process according to these rules. The resource is marked as requires_api_key=false, meaning no additional API Key is needed. This article introduces its positioning, capabilities, activation methods, typical usage, and applicable boundaries: it is suitable for initial paper screening and cross-comparison. The input must be an academic paper; extraction quality depends on the paper's structure; link retrieval is affected by network conditions. Before installation, users should visit the official page to confirm the current installation method and maintenance status.
Read MoreDSH CodeArts Connect: Using Huawei Cloud CodeArts models in DSH
dsh-codearts-connect is a DSH model inference plugin released by maintainer 2239058652. It facilitates Huawei Cloud login via OAuth2 PKCE/DPoP through the settings card and registers CodeArts as a codearts provider. The plugin supports automatic multi-account rotation to mitigate the 3-concurrent session limit per account, hourly model discovery with enabledModels filtering, a local usage ledger, and CLI commands such as status/models/doctor. Installation requires building with pnpm followed by executing deploy.mjs to deploy to the profile. This document summarizes installation, usage, and known limitations based on the repository's README and package.json version 1.3.0.
Read More