Euler's Seven Bridges Stock Analysis: A Skill for Transforming Mathematical Concepts into Investment Frameworks
The PickEulerBridge Stock Analysis Skill is a fundamental company-based investment analysis Skill located at @user_231765ee/pickeulerbridge. Version 1.0.0, maintained by the community-sourced user_231765ee, is categorized as professional and requires no API Key. It transforms Pick's theorem (discrete reduction), Euler's Seven Bridges Problem (topological transformation), and the concept of dimensionality reduction of degrees of freedom into a four-tier analysis workflow: topological positioning, lattice point screening, driving factor analysis, and reflexivity qualification. It is applicable to fundamental company analysis, financial anomaly screening, industrial chain topological analysis, related party transaction identification, event and announcement interpretation, industry landscape assessment, and cyclical stock analysis.
Read Moreflowleap-methods: 15 Patent Workflow Skills for the FlowLeap Patent Agent
Maintained by flowleap-ai, `flowleap-methods` provides 15 patent operation skills (prior art, FTO, invalidation, infringement comparison, Office Actions, etc.) for FlowLeap Patent AI. Each method is located in the `skills/` directory and is invoked via tool names such as `search_patents` rather than executing CLI commands. The DSH is published to `ctx.skills` as a bundle via `dsh/index.js`. Installation requires `dsh plugin add github:flowleap-ai/flowleap-methods#<commit>` combined with FlowLeap patent tools. MIT License.
Read MorePrompt Searcher: Web-wide Search for Pre-made Image and Video Prompts
Prompt Searcher is a Skill designed for image and video generation scenarios, published by user_fca8734f on SkillHub (Location: @user_fca8734f/prompt-library-searcher, Version: 1.0.0, Source: community, Category: knowledge-management, API key not required). Users describe their image or video generation requirements in Chinese, such as "Find a Midjourney prompt for a cyberpunk city night scene" or "Get a Sora prompt for a product advertisement shot." The Skill analyzes the request (type, target model, style/scene/key elements, parameters), performs a search, scrapes source pages, and extracts the full text of real prompts, parameters, and example output descriptions. It then ranks results by relevance, source authority, completeness, and popularity, presenting "1 top recommendation + 3–4 alternatives." The top recommendation includes the full prompt text (English original + Chinese translation), target model, parameters, source link, and a one-sentence description of its applicability; alternatives are listed one per line (Model · Scene · Keywords · Source). Once a user selects a prompt, the Skill outputs a clean, complete, and directly copyable prompt, and asks if it should be saved to the local prompt library by category. The core rule is to only filter and retrieve existing prompts—never rewriting or synthesizing new ones—and to always provide genuine source links. Covered models include MJ / SD / Flux / Sora / Seedance / Runway / Kling / Jimeng. It is suitable for image and video creators who want to use ready-made, traceable prompts; however, niche requests may not yield existing results. Official URL: https://api.skillhub.cn/user_fca8734f/prompt-library-searcher.
Read Moredsh-webannotate: Click-annotate in the DSH built-in browser and batch assign to conversation
dsh-webannotate is a plugin for the DSH desktop version maintained by FleatherYu. It injects an annotation layer into the built-in Electron webview via executeJavaScript, requiring no changes to the target website’s code. Usage involves clicking an element, writing modification notes, and sending everything; annotations are merged into a single session message accompanied by element screenshots (up to 6 per batch). Installation can be performed with the command dsh plugin --profile desktop add dsh-webannotate or github:FleatherYu/dsh-webannotate, and a restart of DSH is required after installation. It is suitable for UI walkthroughs and agent collaboration. Users should be aware of limitations such as iframes, SPA routing, and screenshot limits.
Read MoreVenture Think Tank: Project Analysis Skill Formalizing 'Understand, Innovate, Finalize Content' as a Standard Workflow
Chuangyezhiku (chuangyezhiku) is a community-sourced Skill categorized under business-ops, version 1.0.0, requiring no additional API Key. It fixes the project analysis into a three-stage workflow: deep understanding → innovation upgrade → article generation. First, it builds an understanding of the project or business model, then provides innovation suggestions covering both AI and non-AI dimensions, and finally generates the article. The workflow includes at least three rounds of review and proofreading to verify that the understanding is accurate and the suggestions are well-grounded. Trigger words include "deep analysis", "business model", "product innovation", and "understand before output"; this Skill is used when the user requests analysis of a project, product, or business model, or explicitly asks to "understand first", "execute step by step", or "do not generate directly". The core problem it solves is that such requests are easily skipped in ordinary conversations, whereas this Skill incorporates the understand-first-then-output approach into the execution path via trigger words. The article also describes the installation entry point and typical usage, and provides the following reminders: the source is community-based, the installation count is 0, and the installation commands and client differences have not been verified, so you should first confirm on the official page and perform a small-scale validation before use.
Read MoreDigital Transformation: A Planning Workbench from Current State Assessment to Phased Implementation
This article introduces the "Digital Transformation" Skill on SkillHub. Released by Kunlun Growth (version 1.1.0), it is categorized under "business-ops," sourced from "enterprise," and requires no API Key. It solidifies enterprise digital transformation planning into a fixed pipeline: Status Diagnosis → Gap Analysis → Blueprint Planning → Priority Ranking → Phased Implementation → Impact Assessment. The Skill embeds four methodologies—the Four-Layer Diagnosis Method, Repeated Data Entry Localization Method, "Process First, System Second" principle, and the Six-Month Results Principle—and provides six capability modules corresponding to the pipeline steps. The outputs are directly deliverable, and the entire process includes bottleneck troubleshooting and compliance guardrails. The article explains the specific problems solved by both the pipeline and the methodologies, provides a minimal viable workflow example organized according to the official pipeline, and outlines applicable scenarios and precautions: it is a planning workbench rather than a system implementation tool, and its effectiveness depends on the quality of the current status information. The public page shows zero installs. Note that there is a discrepancy between the ratings and download counts on the official page and other clues; this article uses the data scraped from the official page (rating: 100000, downloads: 978) as the standard.
Read Moredsh-video-eyes: Enabling Pure-Text Agents to Understand Video
dsh-video-eyes is a DSH plugin maintained by fj7937290-sys and categorized under model inference. It uses read_video and read_video_transcript to convert local videos or pure audio into a frame-sampled collage with real timecode, timestamped transcripts, and optional per-segment visual descriptions, without relying on vision-router. The web UI provides a 0.1–10 second frame extraction interval slider; the repository also includes a copy-paste ready Node + ffmpeg skill. Installation command: dsh plugin --profile web add github:fj7937290-sys/dsh-video-eyes. Before use, you must configure a writable artifactDir, ffmpeg, and a Tingbrain Key, and be aware that the plugin runs with the permissions of the Host process.
Read MoreSimulation Experiments for Junior High School Chemistry: Delivering 9th-Grade Chemistry Knowledge Directly to Interactive Simulation Pages
This article introduces the Skill "Simulated Experiments - Junior High Chemistry" (@yimubiao/junior-chemistry-sim), published by Yi Mubiao, version 1.0.2, categorized as education, sourced from enterprise, with the tag requires_api_key=false, requiring no API Key. The skill targets junior high (ninth grade) chemistry, covering knowledge points including a virtual chemistry lab bench, 3D molecular model library, atomic structure, molecular construction, isotopes, the three states of matter, equation balancing, solution concentration, pH basics, and provides Chinese synonyms and English introductions. When a knowledge point is matched, it outputs a hyperlink with a ?kp= parameter that opens the simulation page online; when not matched, it outputs the skill homepage link. The skill is only responsible for providing links and does not render simulation content. The article explains coverage, output branches, installation and activation prerequisites, and provides reproducible usage examples for matched and unmatched cases. Since the official page does not provide curl/npm installation commands, this article does not fabricate them; differences among Cursor, Codex CLI, and Claude Code are not verified and thus not written. Applicable scenarios include junior high chemistry Q&A, review sessions, Agent tool integration, and student self-directed exploration. Notes include: requires internet connectivity to open, matching depends on the model and vocabulary list, does not replace experimental operations and safety protocols, and statistics are based on the official page. The article concludes with the official address and the SkillHub page.
Read MoreCLI Performance Benchmarking Skill: Warm-up, Outliers, and Multi-format Export
Addressing the pain points of CLI benchmarking, the command-line performance evaluation Skill (@user_a10ab3ad/hyperfine, version 1.0.10, source: community, category: dev-programming) offers three verified capabilities: warm-up runs, outlier detection, and multi-format export. This article outlines its positioning, core features, enablement paths, and typical logical workflows. It also details limitations, such as its community source, and notes that the official page lacks installation commands and reproducible examples.
Read MoreObservability and Instrumentation: Making Production Behavior Visible and Diagnosable
Published and maintained by Kunlun Growth, with the identifier @kunlungrowth/observability-and-instrumentation (Version 1.0.0, Source: enterprise, Category: it-ops-security, requires_api_key=false). This Skill targets features that are already deployed but whose online behavior remains opaque. It provides four core capabilities: structured logging, RED/USE metrics, correlation ID tracking, and symptom-based alerting, aiming to make live system behavior visible and diagnosable. The article begins by addressing troubleshooting pain points and is structured in the following sequence: "What it is — Core features — Installation and enablement — Typical use cases — Applicable scenarios and precautions — Conclusion". It accurately states that public documentation lacks installation commands and details regarding platform-specific enablement differences, recommending verification on the official page prior to implementation.
Read Moredsh-supermemory: DSH Cross-Session Long-term Memory (Powered by Supermemory Framework)
dsh-supermemory is a DSH memory plugin maintained by fish9988, with cloud storage and retrieval provided by Supermemory. It performs on-demand semantic recall and hybrid retrieval of historical slices at the agent/pre-step stage, silently refines facts via add/update/delete operations in the background after a quiet interval, applies dual-layer deduplication using Bigram and vectors before writing, and integrates with the Supermemory long-term memory management page. Installation supports GUI paste of github:fish9988/dsh-supermemory or CLI profile commands, requiring an API Key starting with sm_. It provides memory_save/search/forget tools with rich recall, organization, and scope settings, including sanitization and fail-open design. Licensed under MIT, current version 0.1.4.
Read MoreConvertible Bonds & Underlying Stock Research: A Comprehensive Corporate Research Workflow Using Convertibles as the Entry Point
This skill is maintained by community publisher user_fa9720a8 (v1.0.1). It is categorized as Professional, sourced from Community, and tagged requires_api_key=true. It is positioned as a comprehensive corporate research workflow starting from convertible bonds, covering the full chain from convertible bond screening, in-depth analysis of individual bonds, evaluation of down-revision negotiations, underlying equity business analysis, industry comparables, industry cycle, and underlying equity valuation, to an investment summary. Related keywords include conversion premium rate, pure bond value, implied volatility, and historical percentiles of P/E and P/B. According to the official description, the research workflow begins with screening, followed by single-bond analysis and assessment of down-revision negotiations, then business, industry, and cycle analysis of the underlying stock, valuation of the underlying stock, and finally an aggregated investment summary. The official sources do not provide installation commands or specific API examples; please refer to the SkillHub page and API address for details, and prepare your own API key.
Read Moredsh-firefly-pet: Transparent Firefly Desktop Pet for DSH
dsh-firefly-pet is a DSH desktop plugin maintained by Firefly26710Lover. On Windows, it displays a "flying firefly" GIF in a transparent, always-on-top small window via Electron. Chat uses ctx.llm without requiring an API Key, and memory is stored in ~/.dsh/dsh-firefly-pet/memory.json. It supports single-click actions, double-click chat, muttering every 5 minutes, and task announcements triggered by monitoring conversation events. Installation command: dsh plugin --profile desktop add dsh-firefly-pet. See config.json and idle/click-gifs.json for configuration. Windows only; requires DSH Desktop and ~/.dsh/electron. Licensed under MIT, version 0.1.0.
Read MoreRouting Authorized Reverse Engineering and Security Analysis Tasks to the Correct Agent Playbook: An Agent-Oriented Routing Package
This is an agent-oriented routing package Skill, published by community developer user_3c6cb52e, version 1.0.0, categorized under it-ops-security, and based on the general SKILL.md format. It addresses the problem of general agents lacking clarity on what checks to perform first, what tools are available, and what structure to follow when handling authorized reverse engineering and security analysis tasks such as APKs, binary files, mobile devices, APIs, malware, CTF challenges, and penetration testing. It has three core capabilities: routing to corresponding agent playbooks based on task type; performing tool checks before execution; and organizing subsequent steps according to a case structure. For installation, the retrieved materials do not provide an official installation command; it needs to be loaded via the general method of the agent tool being used. The coordinate is @user_3c6cb52e/route-authorized-reverse-engineering-and-security-analysis-tasks-to-the-right-agent-playbook, and no API Key is required. Typical usage covers seven types of authorized tasks: APK analysis, binary file analysis, mobile device analysis, API analysis, malware analysis, CTF, and penetration testing. Each step first undergoes tool checks and case structuring before proceeding to analysis. Notes include: for authorized tasks only, does not include specific tools, the current version is relatively new with zero installations, and it is a non-official product from the community.
Read MoreFund Risk Analysis: Quantitative Risk Analysis Skill for Public Funds
Fund Risk Analysis (@user_d196d58d/investoday-fund-risk-analysis-v2) is a quantitative risk analysis Skill for mutual funds. Maintained and published by community developer user_d196d58d, it is classified as "professional" with a version of 1.3.0. Built upon the Invest Today financial data interface, it automatically identifies fund codes and focuses on metrics such as drawdown, volatility, downside risk, VaR, Beta, Sharpe ratio, and risk-return matching degree to generate structured risk analysis reports. Trigger keywords include fund risk, drawdown, volatility, downside risk, VaR, Beta, Sharpe, and risk-return ratio. The tag indicates requires_api_key=false, meaning no separate API Key configuration is needed. This article introduces its positioning, core features, installation access, and applicable scenarios, while clarifying that installation commands and client differences are not provided in the official documentation.
Read Moredsh-skip-preview: use modifier key to skip DSH sidebar preview and open the file directly with the system default application
In DSH conversations, clicking a file defaults to opening the preview on the right. dsh-skip-preview is a client plugin maintained by finalrgy: on macOS use ⌘+left click, on Windows/Linux use Ctrl+left click. It wraps sidebarRight.openResource to skip the preview and open the file via the system file association; normal clicks remain unchanged. It covers links in the body, inline paths, and the right-hand file tree, but not present delivery cards. Installation: dsh plugin --profile desktop add dsh-skip-preview. A full restart of DSH is required. Requires DSH >= 0.2.1-rc.5. On failure, it falls back to the preview and writes to the Console. MIT licensed; review the source before installing.
Read MoreTraffic Accident Compensation Calculator: Hierarchical Calculation Based on Compulsory Insurance Limits and Liability Ratios
SkillHub Community "Life-Service" Category Skill, Version 1.0.4. Supports Alipay AI Pay-As-You-Go (A2M / HTTP 402); billing settlement is required before invocation, and no API Key is needed. By inputting injury details, disability grade, days of lost work and nursing care, dependents, and accident liability ratios, this skill itemizes compensation calculations including medical expenses, lost wages, nursing fees, nutrition fees, hospital meal allowances, disability compensation, living expenses for dependents, compensation for mental distress, and follow-up treatment costs. It performs tiered settlement of the actual payable amount based on Compulsory Traffic Accident Liability Insurance limits and liability ratios, and generates a downloadable, printable calculation report for claims negotiation and litigation preparation. The official trigger words cover Chinese and English colloquialisms as well as technical terminology; activation immediately provides itemized breakdowns, final amounts, and reports. Note: Disability grades and liability ratios are subject to forensic appraisal conclusions and accident liability determination letters. Calculation results do not constitute legal advice.
Read MoreScript Extraction from Douyin, Bilibili, and Kuaishou Videos: Offline ASR Transcription for Videos Without Subtitles
This article introduces the "Douyin, Bilibili, and Kuaishou Video Voiceover Transcription" tool on SkillHub. By inputting Douyin share text, Bilibili BV numbers, or other video links and descriptions, the tool downloads audio streams and utilizes the offline Vosk speech recognition engine to perform Chinese transcription. It automatically cleans ASR errors and formats the output. It is suitable for scenarios involving videos without subtitles, the need for complete verbatim transcripts, or processing long videos without using online APIs. No API key is required. The article outlines core features, trigger words, applicable scenarios, and limitations, while truthfully noting that the official page could not be accessed and the installation command could not be verified at the time of review.
Read MoreStorage Tank CAD Drawing Generation: Keyword-Triggered Drawing Creation in Agent
This article introduces the community Skill "Tank CAD Drawing Generation" (@user_5453610a/tank, Version 1.0.0, Category: dev-programming). It addresses the challenge of consolidating the "on-demand generation of tank CAD drawings" into a single Skill entry point. The Skill is triggered when users input commands such as "储罐" (tank), "生成储罐图纸" (generate tank drawings), "做一个储罐" (make a tank), or "tank," subsequently outputting the CAD drawings. Tagged with `requires_api_key=false`, the Skill requires no additional keys for invocation. As the SkillHub page does not disclose installation commands, directory conventions, or configuration fields, nor does it detail the enablement differences for Cursor, Codex CLI, and Claude Code, this article strictly documents verified location coordinates, source permissions, and trigger mechanisms. It makes no speculation regarding CAD output formats or configurable parameters; it is recommended to defer to the actual behavior observed after activation. The source is community-based, and it is advisable to review the publisher's content before use. The page ratings and download statistics are snapshots and subject to change over time.
Read Moredsh-web-terminal: Displaying a Real PTY Terminal in the Right Sidebar of the DSH Web UI
dsh-web-terminal is a client plugin for the DSH web profile that imports the agent's foreground shell into a persistent PTY, displays a Claude-style terminal in the right sidebar, and provides 17 live_terminal_* tools. Based on the official README, this article describes the applicable scope (DSH >=0.2.1-alpha.3), the install.ps1 and manual installation steps, verification methods, main configuration options, and security considerations regarding the absence of file sandboxing. It also provides the SkillHub directory and GitHub links.
Read MoreAI-Powered PPT Presentation Video Generator: Transforming PPTs into Scripted Talking Head Videos
This article introduces the PPT Intelligent Speech Video Generator published by Guomeng Digital Technology on SkillHub (coordinates @guomeng/ppt-to-speech-video, version 1.1.0, category office-efficiency). The skill addresses the pain point of manually synchronizing scripts and visuals when converting PPTs into voiceover videos: after uploading a PPT, it automatically parses content, understands logical structure, and generates a speech script with support for 5 different speaking styles. It is a paid skill that requires no external API key configuration. The article verifies both the parsing and script generation capabilities, and notes that the official description is truncated in the speech synthesis section, so specific interfaces and output formats of the synthesis pipeline should be referred to the official page. No installation commands, configuration fields, or usage examples were fabricated. The article also outlines applicable scenarios (generating multi-style scripts from the same source, avoiding separate maintenance of script documents) and considerations (scripts may become vague when PPT content is minimal, AI-generated scripts require human review, and the skill is paid with zero installations).
Read MorePonytail (Minimalist Lazy Development Mode): The Minimal Change Solution for Coding Tasks
Ponytail is from ClawHub, published by dietrichgebert, with a scraped version of 4.8.4. It is categorized under dev-programming and tagged requires_api_key=false. Its official introduction positions the tool to handle any coding task, including writing, refactoring, fixing, and code review. It adheres to the YAGNI principle, prioritizes the standard library, avoids unnecessary abstraction, and explicitly states it is not applicable to non-coding requests. The main body is organized in the order of four core capabilities, installation coordinates, typical usage sequence, applicable scenarios and relevant notes, and a brief conclusion. The ratings and download counts provided in the reference clues are inconsistent with the scraped data and were not adopted. Trigger words mentioned in the clues and the three tiers (lite/full/ultra) do not appear on the official page, and this discrepancy is noted factually as-is.
Read Moredsh-kassel-boot: Kassel College style DSH boot animation
dsh-kassel-boot (v0.1.1, MIT), maintained by Ffwjl010831, plays a six-screen Kassel College-style authorization sequence in shell.overlay during DSH startup. It requires clicks or Enter to advance screen by screen; SKIP jumps to the last screen. The plugin automatically uninstalls after a dark gold transition at the end. For installation, it is recommended to download the tgz file and use 'dsh plugin add'. Alternatively, you can use install.sh or paste the jsDelivr address on the plugin page. Do not install using the github: format.
Read MoreEmo-Script: Establishing Script Skeletons, Before Inputting Inspiration into AI Video Tools
Emo-Script-AIGC is a script development system designed for AIGC short films. Identified as @user_64a1eae5/emoscript (Version 1.0.0), it is a community-sourced tool categorized under content-creation and tagged with requires_api_key=false. Positioned as a precursor to generation tools like Jimeng, Runway, Pika, and Kling, it addresses the issue of "missing narrative arcs caused by direct prompt-to-image generation." It establishes a structural framework before visual generation by analyzing three key aspects: the nature of the emotion, the scene's position within the emotional arc, and the character's trajectory (ascending or descending). It supports two modes: a Generation Mode that expands a single sentence of inspiration into a complete production bible, and a Discussion Mode that provides in-depth analysis of scenes, characters, objects, or pacing. Trigger keywords cover scriptwriting, short film production, storyboarding, story outlines, plot design, emotional rhythm, AIGC prompts, production bibles, world-building, character shaping, and ending resolutions. A fallback rule ensures activation for any story, emotional expression, or visual design related to short videos, films, or AIGC creation. Public information does not provide command-line installation methods; users should refer to the official page for instructions. Current page data shows a rating of 100,000, 982 downloads, and 0 installs; adoption remains low, and functional performance requires independent verification.
Read Morefollow-builders-lite: A lightweight AI builders daily report reading only public feed JSON
Follow-builders-lite is a lightweight Agent Skill derived from the upstream zarazhangrui/follow-builders project. Hosted on SkillHub, it is categorized under "knowledge-management" with the tag `requires_api_key=false`. This skill directly reads public feed JSON files (feed-x.json, feed-podcasts.json, feed-blogs.json) generated by upstream GitHub Actions to compile summaries of AI builders' activities. It requires no API key, cron jobs, delivery channels, or local scraping scripts. It is ideal for generating on-demand summaries within conversations but does not support automated delivery. This article introduces its positioning, feed structure, typical usage, and applicable limitations.
Read More