Dual-Lens Conference Advisor: Conducting a "Hard Metrics + Relationship Insights" Analysis
Shuangjing Conference Advisor is a universal SKILL.md-format skill released by user_c2013bda in the SkillHub community, classified under office-efficiency (v1.0.0). It does not require an API Key. It scrutinizes evidence, cost, accountability, and acceptance criteria via Hu Chenfeng-style hard metrics, and handles stakeholder interests, resistance, face-saving concerns, and phrasing for progress via Tong Jincheng-style relational insights; it covers pre-meeting evaluation of meeting value, in-meeting detection of empty rhetoric and pseudo-consensus, and post-meeting distinction between factual assessment statements and genuine decisions. Outputs: Double Mirror Index, Consensus Quadrants, Rhetoric Tax, Commitment Fingerprint, Decision Ledger, Action Contract, and follow-up phrasing in three versions — direct, graceful, and executive. It supports pasting topics, meeting minutes, or verbatim transcripts, and supports reading local TXT, Markdown, and JSON files; conditionally triggers when the official Tencent Meeting tmeet Skill is detected for a read-only retrieval of Smart Minutes and AI Transcriptions. It also functions offline without Tencent Meeting. It is ideal for decision meetings, product reviews, project retrospectives, weekly syncs, client meetings, management meetings, and proposal discussions, but not suited for casual chats, romantic advice, voice mimicry, or unrelated product reviews.
Read MoreTheme Stylist: Skill for Unifying Color Schemes in Slides, Reports, and HTML Pages
主题样式师 (Theme Stylist) is a theme-styling design Skill, released and maintained by 智创未来 (ZhiChuang Future), version 1.0.0. It is sourced from the enterprise channel and categorized under design-media. By applying theme customization, color schemes, and consistent typography, it drives slide beautification, document layout, report color coordination, and the creation of HTML landing pages, reducing repetitive manual color adjustments for each document. API keys are not required (requires_api_key=false). Official installation commands and invocation formats are not provided in existing materials; please refer to the official address page for details.
Read MoreTopic Selection and Title: Introduction to the WeChat Official Account Topic and Title Generation Skill
"Topic Selection & Title Creation" is an AI Skill designed for official account (WeChat) editors and content operators, covering five key stages: trending topic research, topic planning, title generation, summary writing, and series scheduling. Published by RealQ, this skill is sourced from "enterprise," categorized under "content-creation," and is currently at version 1.0.0. It holds a perfect SkillHub rating of 100,000 with 1,042 downloads, and its tags indicate that no API key is required. Please note that trigger words are strictly for modifying existing titles or summaries (e.g., "change title," "optimize title," "change summary"). For creating new topics, generating new titles, planning content series, or chasing trends, please utilize the main skill "aws-wechat-article-main." This main skill also supports串联 (chaining) multiple workflow stages. This article introduces the core functionalities, trigger word boundaries, activation methods, and applicable scenarios.
Read Moresapi-budget: Sync AiSpot Grouped Pricing in DSH and Display Session and Sub-Agent Costs
sapi-budget is a DSH model inference plugin maintained by Lanstzz. It fetches data from `/api/pricing`, `/api/status`, and `/api/user/groups` in AiSpot (sapi) compatible deployments. It displays the total conversation cost—including the subagent subtree—in the conversation title, and shows the cost per turn at the end of each assistant message, with hover support for viewing token breakdowns. User groups are dynamically synced from the provider, and switching groups in the UI triggers immediate recalculation of historical data. To install, run `dsh plugin add github:Lanstzz/sapi-budget` followed by a restart. Optional `cordis.patch.yml` configuration is available for `baseUrl`, `currency`, and `refreshMinutes`. Costs for hierarchical models and subagent attribution are marked as estimates, and subagents that terminated before the plugin was enabled are marked as "unknown". Licensed under MIT.
Read MoreSteering Gear CAD Drawing Generation: Delegating Drawing Requirements to Skill
**Steering Gear CAD Drawing Generation** (@user_5453610a/steering-gear) is a community Skill published by user_5453610a on SkillHub. Version 1.0.1, it falls under the design-media category and is tagged with requires_api_key=false. The page displays a rating of 100,000, 929 downloads, and 0 installations. Its function is to recognize phrases such as "steering gear," "generate steering gear drawings," "make a steering gear," or "steering-gear" in conversations and trigger the generation of steering gear CAD drawings, transforming the originally manual process of opening CAD software to create drawings into a single natural language command. The article introduces its positioning, core features, trigger methods, applicable scenarios, and precautions: it only covers steering gears; the official documentation does not specify drawing formats, precision standards, or dimensioning conventions; no installation commands or configuration differences across hosts are provided; users should refer to the official page for accurate information. It is suitable for mechanical design Agent workflows that aim to automate the process from "a single sentence to drawings."
Read Moredating-time: Interactive Simulation Agent Skill for the Matchmaking and Dating Process
dating-time is a community-sourced lifestyle service Skill, identified as @user_b1d2120e/dating-time, currently at version 3.0.0. Positioned as "AI Blind Date & Romance Simulator v3," it randomly generates unique virtual matchmaking partners within the Skill, simulating the complete process from adding WeChat contacts to establishing a romantic relationship, while supporting pause/save, conversation continuation, and meeting mechanics. Official trigger words include "blind date," "dating simulation," "dating sim," "simulated romance," "date simulation," "blind box date," and "random date." The Skill does not require an external API Key (requires_api_key=false). The SkillHub page displays a rating of 100000, 908 downloads, and 7 installations. It is suitable for testing Agent role-playing capabilities, generating dialogue-based content assets, and validating the resumption of saved states in long-running interactions. This retrieval attempt was unable to obtain specific installation commands, directory conventions, or configuration file fields; relevant steps have not been fabricated in this text. For actual usage, please refer to the installation instructions provided on the official page.
Read MoreHandoff-Skill: Compress the Current Conversation into a Handover Document
When handing over long conversations, simply transferring the context forces the receiving party to dig through noise for conclusions. The "handoff-skill" is a community Skill (Community Skill @user_a613e9fd/handoff-skill, Version 1.0.0, Category: ai-agent) described officially as compressing the current conversation into a handoff document, allowing another agent to take over directly. The page tag `requires_api_key=false` indicates that no additional key configuration is needed for activation. Since the official page does not provide installation commands, directory paths, or output formats, this article only explains the invocation intent and verification methods without reproducing any third-party commands. It is suitable for scenarios where the unfinished work of a single agent needs to be handed over to another, and the original conversation has become excessively long. Given that the introduction contains only a single sentence ("test handoff skill") and the functional description is merely one line, it is recommended to first conduct a small-scale verification to ensure the compression results retain goals, verified conclusions, and pending tasks before putting it into actual use.
Read Moredsh-network-proxy-plus: Network Proxy and SOCKS5 Configuration for DSH 0.2.x
An introduction to the DSH plugin *dsh-network-proxy-plus* maintained by Lanlan-gys: it serves as a replacement for the legacy *dsh-network-proxy* that fails to load in version 0.2.x. It allows configuration of SOCKS5, SOCKS5H, HTTP, and HTTPS protocols via the settings page; changes to the 'volatile' field take effect immediately upon saving, and the proxy environment is synchronized for child processes. The documentation includes the GitHub installation source, the rationale for defaulting to system settings, loopback direct connections, and known limitations.
Read MoreEducational AI Assistant: Codifying Teaching Scenario Prompts into a Triggerable Skill
The Teaching AI Assistant (@kunlungrowth/kg-edu60-009), released by Kunlun Growth (version 1.0.0), is classified under the education category. Its tag indicates that no API key is required (requires_api_key=false). The core mechanism relies on trigger words—teaching, academic advancement, education, and similar terms—and upon detecting these trigger words, it processes conversations according to educational scenarios. The article outlines its positioning, verified core capabilities, and the method to enable it based on SKILL.md. It also transparently notes that the official page was inaccessible during this review, and no installation commands or configuration fields were retrieved.
Read MoreTax Expert Group on Capital Operation and Restructuring: Consolidating M&A and Restructuring Tax Issues into a Low-code Skill
The Capital Operations & Restructuring Tax Expert Group is a general-purpose Agent Skill in SKILL.md format, located at @user_11064e10/tax-expert-capital-restructuring. Version 3.18.64. Source: community. Category: professional. No additional API Key required. It is coordinated by a Capital Tax Director and integrates domain specialists covering M&A restructuring, capital reduction and exit, equity governance, employee stock ownership plans (ESOPs), and IPO review compliance. It provides tax planning and compliance implementation support across the entire lifecycle of corporate capital operations. The article describes its positioning in the writing style, verified core capabilities, official entry point, and activation method (no installation command is provided in public materials, so it is not quoted verbatim). Based on the brief, typical use cases, applicable scenarios, and precautions can be anticipated. It also notes that tax conclusions should be reviewed by qualified professional advisors.
Read MoreComic-style Short Drama Videos: Integrating Generation into AI Workflows
Introduce the comic drama video Skill on SkillHub (Handle: @infimind/comic-drama-video, v1.0.0, category: design-media; publisher: Jray Tech - AI Expert). Acting as a general entry for searching across commercial scenarios, the Skill calls AI Hive’s raw APIs under the hood, supporting needs such as comic drama video, AI comic drama, comic short drama, vertical-scroll comic video, comic series, 2D short drama, Seedance, MiniMax H3, Happy Horse, and AI video. The tag requires_api_key=true indicates that an API key is required to run. The article is organized as follows: Preface, Positioning, Core Features, Installation and Activation, Typical Usage, Applicable Scenarios and Notes, and Conclusion. Only verified facts are included; sections where installation commands, configuration fields, or reproducible call examples are not provided are explicitly left blank. Unverified device-side differences are not covered.
Read Moredsh-yinji: Selective Hiding of Workspace and File Tree in DSH
隐几 (Yǐnjī) is a DSH 0.2.x client plugin maintained by Landau1994. It hides lists and sessions by workspace in the left sidebar, and hides directories and files in the right panel via the files tab extension. The hide list is stored in browser localStorage and persists after a refresh. It only affects the UI; the model's read/glob/grep and session data are unaffected. To install, run `dsh plugin --profile desktop add <local-path>` and restart Desktop. MIT licensed.
Read MoreHomework Tracker: Turning Class Group Chat Photos into a Checklist
Introduction to the community Skill "Homework Tracker": It automatically extracts homework information from assignment photos in class WeChat groups, organizes them into an actionable checklist, and writes it into Tencent Docs. It is triggered when users send photos of handwritten assignments, forward assignment images, or use terms such as "作业", "homework", "更新作业", or "记录作业". The Skill is version v1.0.0, coordinates @user_730c93d7/homework-tracker, sourced from community, categorized under life-service, and does not require an API key. The article describes its core functions, triggering methods, applicable scenarios and limitations, and provides the official link.
Read MoreAssistant for Annual Review and Promotion Defense: Organizing Scattered Work Records into Structured Reports
The Annual Performance Review and Promotion Defense Assistant is a dual-mode workplace tool released by community publisher user_fa14b5bc, version 1.0.0, categorized under office-efficiency, with coordinates @user_fa14b5bc/performance-review-assistant, and labeled requires_api_key=false. The performance review mode uses a three-step guided information collection process to transform scattered work records into a structured report under the STAR framework, comprising five sections: annual overview, key work review, capability growth summary, shortcomings and improvements, and plans for the coming year. The defense mode provides promotion qualification self-assessment, defense material optimization, predictive question bank generation, and mock defense scoring, evaluating across three dimensions—logical clarity, data support, and conciseness of expression—while offering optimized phrasing suggestions. It incorporates 9 categories of content compliance review rules to ensure outputs conform to workplace-standard expressions. Since the materials do not provide installation commands or official reproducible examples, and the official page is currently inaccessible, the text only describes the features and usage paths without fabricating installation steps.
Read Moremesh-memory: Cross-device Agent Shared Memory Based on Markdown and Syncthing
mesh-memory (LanceRingHong) is a memory plugin for DeepSeek Harness: with Markdown as the source of truth, it supports multiple entry points via DSH, MCP, CLI, and HTTP, Syncthing for directory synchronization, and cross-device task delegation through DSH Remote. Memory entries feature checksum-based optimistic locking and lifecycle management, automatic BM25 retrieval and a session-summary inbox, while maintenance skills handle reflection and organization. Installation is performed via the repository's install.ps1 script writing to the profile, rather than `dsh plugin add`. MIT licensed; requires Node ≥ 18 and DSH 0.2.x.
Read MoreLong/Short Judge: A Skill for Long-term and Short-term Diagnosis of 6-digit A-share Stock Codes
The Long/Short Term Judge is an Agent Skill published by SkillHub community user user_6c1a7171 (Version 1.0.0, Category: Professional, Source: Community). It is triggered when users provide a 6-digit A-share code and inquire: "Is this stock suitable for long-term or short-term investment?", "Is it suitable for long-term holding?", "Can I hold it long-term?", "Is it suitable for short-term trading?", "Should I go long-term or short-term?", or "Help me diagnose this stock." Upon activation, the skill retrieves real-time market quotes, daily technical indicators, historical valuation percentiles, and financial data as well as goodwill across multiple reporting periods. It performs dual scoring based on five long-term dimensions plus five short-term dimensions, outputting a comprehensive diagnostic report (including operational recommendations) consistent with the original web version. Trigger words: Long/short judgment, Long/short term judge, Suitable for long-term or short-term, Long-term or short-term, Suitable for long-term holding, Worth holding long-term, Suitable for short-term trading, Individual stock diagnosis. Tag: `requires_api_key=false` (No additional API Key application is required before use). Installation: Place the files into the skills directory of Cursor, Claude Code, or Codex CLI following the standard SKILL.md format. No explicit loading in the conversation is necessary. Note: Only supports 6-digit A-share codes. Diagnosis relies on networked data and the output does not constitute investment advice.
Read MoreRecruitment JD and Interview Question Bank Generator: Delegating JD Writing and Question Generation to a Skill
The JD & Interview Question Bank Generator (@douease/hr-jd-interview-kit) is an end-to-end recruitment Skill released by the DouEase team. As a Version 1.0.0 tool categorized under business-ops, it requires no API key. It covers over 20 recruitment requirements, including JD writing, job profiling, competency models, interview question banks, structured interviews, STAR follow-up questions, interview evaluation forms, and salary negotiation. It features a built-in three-step job profiling method, structured interview question banks and follow-up question chains across five competency dimensions, and evaluation forms with scoring anchors. The article provides installation commands, typical trigger prompts, and applicable scenarios.
Read MoreQingFlow MCP: Invoking Business Data in AI Agents
Qingflow MCP is an MCP skill tailored for AI agents, published by community developer user_2b6572ff. It is categorized under business-ops, tagged with community as its source, and is currently at version 1.0.0. It addresses a common pain point: when developers use Qingflow to manage business processes, data remains trapped on the server. Performing analysis or automation often requires exporting data for manual processing, or writing custom authentication and API request code just to access the endpoints. Once integrated, users simply need to mention Qingflow to allow the agent to log in and directly invoke Qingflow data for analysis, or manipulate business data. It is positioned as a bidirectional capability tool for both analysis and operation, rather than merely a read-only reporting dashboard. The official description states that mentioning Qingflow serves as the trigger condition. Although the page lists requires_api_key=false, it also mentions that usage requires a login, so actual authentication requirements need to be verified independently. The documentation does not provide standalone installation commands or configuration fields, nor does it list the specific data tables and APIs available for invocation. Installation must be performed via the platform's standard workflow from the SkillHub skill page, and the exact capability boundaries should be based on the official page descriptions. It is highly recommended to validate it in a testing environment before making a final selection.
Read Moredsh-balance-simple: DSH Sidebar DeepSeek Balance and Peak/Off-Peak Status Panel
dsh-balance-simple is a DSH Web plugin (v1.0.2, MIT) maintained by lamp23333. It displays the DeepSeek balance, monthly usage, costs, cache hit rate, and Beijing time peak/off-peak status along with a switch countdown above the sidebar avatar. A 44px status block remains visible even when the sidebar is collapsed. It reuses existing DSH DeepSeek authorization to fetch data via same-origin interfaces and features multi-level fallbacks for holidays and NTP. To install, run `dsh plugin --profile web add dsh-balance-simple` or use the git+GitHub address, then simply refresh the page.
Read MoreTavily MCP Server: Real-Time Web Search and Page Extraction for AI Agents
The Tavily official MCP server provides four capabilities for AI agents: real-time web search, page extraction, site mapping, and web crawling. It connects to the Tavily Search API to return structured results optimized for LLM consumption. This Skill is labeled as originating from the "Community," published by user_3c6cb52e, and is at version 1.0.0. It falls under the "AI Agent" category and carries the `requires_api_key=true` tag, meaning a Tavily API Key must be prepared before calling. The page displays a rating of 100,000, 1,023 downloads, and 0 installations; given the abnormal rating magnitude and zero installation count, these figures should not be used as a basis for quality assessment. During this writing, both the SkillHub page and the Tavily documentation failed to be retrieved. Therefore, installation commands, MCP configuration fields, configuration differences for Cursor / Codex CLI / Claude Code, and typical call examples were not included in the text. It is necessary to verify this information at the official address before integration.
Read Moredsh-spend-pill: Display actual cost of last query in Composer status bar
dsh-spend-pill is a DSH client plugin maintained by lalilulelo3 (v0.1.1, MIT) that displays the actual settled cost (not an estimate) of the previous sentence in the Composer status line. Clicking it reveals the cumulative session total, token breakdown, model call count for that turn, the current in-progress cost, and the starting price for re-reading history in the next sentence. Based on a spendStats projection of the full session log, it ensures that restarts and context compression do not affect the settled values; it also indicates the remaining context before compression at 50%/60% thresholds. To install, run `dsh plugin --profile desktop add "github:lalilulelo3/dsh-spend-pill"` and restart the host. Pricing is hardcoded according to official DeepSeek peak and off-peak rates; this is not an invoice, so discrepancies may occur during mixed-rate periods.
Read MoreCreating Editable HTML Courseware: Turning Materials into Single-file Interactive Web Pages
This Skill transforms user-provided images, files, textbook screenshots, and text into single-file HTML interactive courseware. The output is instantly usable—requiring no installation and running like an app with a simple double-click. During generation, it supports reserving placeholders for images, charts, and videos at appropriate locations based on prompts. The layout automatically adapts to the source materials with reasonable sizing, ensuring no overflow or obstruction. It features a left-side directory navigation bar and a back button. This tool is ideal for creating courseware, generating presentations, turning materials into web pages, creating educational slides, converting PPTs or screenshots into paginated web pages, and building interactive pages for textbook content. **Publisher:** user_3723f934 **Version:** 1.0.0 **Source:** Community **Category:** Education **API Key:** Not required The article covers core features, installation access, typical use cases, and precautions.
Read MoreGuide to OpenAI Official Documentation (OpenAI Developer Docs): Check Official Docs Before Responding
A command-line Skill for querying OpenAI's official developer documentation. Published by am-will, sourced from clawhub, currently at version v1.0.0, licensed under MIT-0, with a Pass security audit, SkillHub rating of 100000, 6,095 downloads, 130 installs, categorized under ai-agent, requiring no API key. The core is a shell script located at scripts/openai-docs-mcp.sh, which uses curl + jq at its base to query the OpenAI Docs MCP server at https://developers.openai.com/mcp, providing seven subcommands: init, tools, search, list, fetch, endpoints, and openapi. The MCP_URL environment variable allows overriding the endpoint. The officially recommended workflow follows three steps: search (use list when uncertain) → fetch → cite URL. Installation via `openclaw skills install @am-will/openai-docs`. It is suitable for scenarios requiring the latest official documentation on the OpenAI API (Responses, Chat Completions, Realtime), SDKs, ChatGPT Apps SDK, Codex, MCP integration, endpoint structures and parameters, rate limits, and API migrations.
Read MoreWeipu Writing: Integrating Literature Retrieval, Paper Structuring, and Drawing into a Single Academic Writing Skill
VIP Writing (@vipsmart/vip-writing), released by VIP Data, originates from the 'enterprise' source, falls under the 'education' category, and is version 1.0.1; an API Key is required for activation. Backed by 1.2 billion global document resources and billion-level full-text resources, it targets journal papers, review papers, conference papers, graduation theses, literature review, and scheme design. It incorporates writing techniques for various disciplines and diverse charting skills, supporting both Standard and Tutoring modes. Since publicly available data does not provide directly verifiable installation commands, directory paths, or configuration fields, this article focuses solely on verified coordinates, versions, sources, and capability scopes. It cautions users against copying key configurations from other Skills or adopting inconsistent rating and download data. This tool is suitable for scenarios requiring an integrated workflow for literature research, long-form drafting, and illustration creation.
Read Moredsh-sidebar-quota-widget: Sidebar Quota Ring and Official Balance
The dsh-sidebar-quota-widget (v1.0.0, MIT) by lakii2982 displays the official DeepSeek balance in the Web GUI sidebar account row, utilizing a dual-ring visualization to show 5-hour and 7-day remaining quotas for locally connected providers such as OpenCode Go, Kimi Coding, and Zhipu. The Host reads data from real usage APIs with a 45-second cache, while the frontend polls every 60 seconds; API keys are resolved via DSH credentials and remain hidden from the frontend. Installation is performed by cloning the repository and either using the Plugin Manager for local bundling or using the absolute path with install_bundle. The widget auto-hides when the sidebar is collapsed, and displays a gray ring for unsupported providers.
Read More