Project Management for Non-Professional Project Managers – Practical AI Project Management: Delegating Project Management Execution to AI
Released on SkillHub by the Baitong AI Service Platform, this Skill is classified under "business-ops" (version 1.0.1) and requires no API Key. Its official description states: "Master the essence of project management through this skill pack!" Designed as a teaching and practical guidance tool, it encapsulates project management methodologies into AI-callable workflows. It is tailored for developers and small teams lacking a dedicated PM who seek to implement a basic methodology for their projects. The article outlines its positioning, verifiable features, activation methods based on the host tool, typical use cases, and precautions, providing an official link for verification.
Read Moredsh-session-tariff: Display estimated cost for current session in DSH input status line
dsh-session-tariff is a DSH client plugin maintained by xuanyuanli. It displays a pill showing the estimated cost of the current session in the input box's status bar, with pricing based on peak/off-peak rate cards from providers such as deepseek-account and Bailian. The host's sessionCost projection collapses session events, while clicking it reveals the token bucket breakdown by time period and model. It supports editing pricing rates on the settings page with hot updates and is compatible with @deepseek-ai/dsh ^0.2.1-rc.2. To install, run `dsh plugin add dsh-session-tariff` or use the assistant's `install_bundle` in the source directory.
Read MoreMedical GAP Analysis Skills: Advancing from "Problem Identification" to Evidence Generation and MSL Implementation
Introducing the “Medical GAP Analysis Skill” (`@baitongai/medical-gap-analysis-skill`, v1.0.0, Source: enterprise, Classification: professional) on SkillHub. Based on the STEPS framework, it considers the perspective of a senior medical strategy manager; diagnoses product gaps across three dimensions—evidence, cognition, and treatment pathways—and provides a closed-loop solution based on the three-pillar pipeline of GAP analysis → evidence generation plan → MSL deployment and integration. It addresses the pain point in medical strategy of uncovering problems without resolving them. This article explains its capability limits, output formats, official metadata (rating 100000, 762 downloads, requires_api_key=false), and usage restrictions. It should be noted that the official page does not provide install commands, third-party adapters, or reproducible call examples; we have not fabricated these, and you need to check the official SkillHub page for verification.
Read Moredsh-ziwei-doushu: Open Local Zi Wei Dou Shu Chart Casting in DeepSeek Harness
dsh-ziwei-doushu is a companion plugin for the DSH client (v2.0.1, MIT), maintained by XuanCenXuKe. It integrates the locally deployed ziwei-doushu application into the Harness via a sidebar iframe, retaining full functionality including astrological chart casting, the Four Transformations, Major Life Cycles, compatibility charts, ancient texts, and AI-driven destiny readings. The Host features semi-default probing every 30 seconds and automatically executes 'next start' if the application is not running; the plugin does not access API keys. Installation can be performed via 'install-ziwei.sh' to build the application or through 'plugin_manager install_bundle' from GitHub. Optional configurations include ZIWEI_APP_DIR, ZIWEI_NODE_BIN, and ZIWEI_AUTO_START. This solution is ideal for users seeking to utilize Zi Wei Dou Shu in a local closed-loop environment while simultaneously collaborating with Agent workflows on the same screen.
Read MoreThe Chinese Obsession with Reading Between the Lines: Teaching Agents to Decode Chinese Subtext
This is a deep reading comprehension and subtext analysis Skill designed for the Chinese linguistic context, identified as @user_06be1fa4/deep-reading-cn, Version 1.0.0. Maintained by community publisher user_06be1fa4 and categorized under education, it requires no API Key. It turns the analysis of four categories—subtext, underlying emotional currents, implicit stances, and author motivation—into a fixed workflow. It is applicable to chat records, Weibo, WeChat Official Accounts, Xiaohongshu, Tieba, Zhihu, and the Chinese Twittersphere. It supports both single-article analysis and comparative interpretation of multiple texts, allowing users to either paste text directly or provide a URL for automatic retrieval. This document introduces its positioning, capabilities, activation methods, and typical usage, while clarifying that the Skill's results are inferential in nature and do not replace fact-checking.
Read MoreDouyin Search Analysis: A Skill for Public Information Research
Douyin Search & Analysis (@user_7a394f02/douyin-search-and-analyze-guaikei) is a community-published business-ops Skill (version 1.0.0) designed for researching public information on Douyin. Its four core capabilities include searching content, scraping creators' works, reading comments, and checking trending searches. The Skill is also applicable when users discuss trending topics or competitor account activities on short-video platforms without explicitly mentioning Douyin. However, it is not designed for video editing, publishing, or downloading. Integration requires an API key configuration (requires_api_key=true). Since the provided documentation lacks specific installation commands and official examples, the text mainly illustrates capabilities and usage directions; any unverified aspects are note-worthy.
Read Moredsh-provider-hub: Align model provider references in settings to eliminate NO_ADAPTER
Maintained by xuanbaon, **dsh-provider-hub** is a DeepSeek Harness plugin that aggregates settings in the "Provider Center" to verify consistency between actual session models, configuration references, and profile definitions. To address the common **NO_ADAPTER** error in headless environments, it supports one-click definition restoration, batch or line-by-line provider switching, pre-write backups, and atomic writes. Installation involves declaring the dependency and bundle in the profile's `package.json`; DSH requires a restart after configuration changes. It also provides a `sync` command-line and `/dsh-provider-hub/*` HTTP interfaces. MIT 1.2.0.
Read MoreValuation and Rating of Enterprise Technology Patents: Grading Patents and Patent Pools per GB/T 42748-2023
Enterprise technology patent evaluation and rating skill, located at @user_07bb1a57/patent-evaluation. Version 1.0.1, source: community, category: data-analysis, API Key not required. Complying with the GB/T 42748-2023 "Guidelines for Patent Evaluation," it performs patent value assessment, patent pool rating, hierarchical classification, high-value patent screening, maintenance/abandonment decision-making, and report generation within the conversation. The scope covers Patent Value Degree (PVD) and patent evaluation system consulting. Upon installation, it is triggered via natural language; inputs include patent bibliographic data and portfolio information, while outputs provide results categorized by value degree. Captured metadata shows a rating of 100,000, 1,192 downloads, and 0 installs. Please note that the completeness of input affects result reliability; formal scenarios still require reports issued by professional institutions.
Read More植物日记：面向花草养护的 Agent 技能介绍
植物日记是社区版 Agent Skill（@user_407ad525/liangyi，v1.0.0，分类 life-service），定位花草养护助手，无需 API Key。能力覆盖植物识别与健康诊断、浇水施肥计划与天气联动、植物日记与档案管理、繁殖指南与购买建议，以及面对发芽开花等开心或植物死亡等难过情绪时的感知与告别流程。触发词包括植物养护、叶子发黄枯萎、病虫害、扦插繁殖、植物档案等。文章介绍其定位、功能、接入方式、典型用法与限制，接入命令需以 SkillHub 官方文档为准。
Read Moredsh-emotion-memory: Providing conversational agents with PAD emotional field and emotion-fingerprinted memories
dsh-emotion-memory is a memory plugin maintained by xsy822 for DeepSeek Harness. It offers a PAD 3D emotion field, 14-channel readout, lightweight per-turn evaluation with derived emotions, and a memory mechanism that attaches emotional fingerprints during writing and performs reconsolidation during retrieval. It injects a "Base Emotion + Recalled Memory" context based on weighted scoring. To install, run `dsh plugin add dsh-emotion-memory`. Configuration and state files are located in `~/.dsh/emotion-memory/`, and memory entries reuse the fish-memory session JSON format. Based on the official README, this article covers installation, configuration, export/import, and known limitations to assist developers in evaluating potential integration.
Read MoreDSH-MCU51: DSH 8051 Microcontroller Toolbox for Windows
dsh-mcu51 is a Windows x64 DSH client plugin (v0.3.0-alpha.3) maintained by XSXLL22. It provides 12 development tools for 8051/STC, including mcu_doctor, mcu_flash, mcu_serial, and mcu_options. Official Releases bundle an isolated Python environment and stcgal, eliminating the need to rely on system-wide Python or STC-ISP. This article covers the scope of tools, installation commands for official Releases, the typical dryRun flashing workflow, as well as host requirements for the DSH 0.2.1 series and verification boundaries for the alpha version.
Read Moregeo-article-writer：Introduction to the GEO Article Generator for WeChat Official Account Customer Acquisition
geo-article-writer is a Skill designed for automated GEO article generation tailored for WeChat Official Account scenarios. Coordinates: @user_1831d4fb/geo-article-writer. Version: 1.0.0. Source: Community. Category: content-creation. Tag: requires_api_key=false. It solidifies the writing formula of "viral titles + storytelling narratives + customer acquisition features" into a WeChat article ready for the editing stage, reducing the cost of drafting from scratch each time. The official introduction also states that the service belongs to the GEO Optimization Classic Manuscript Service provided by Beijing Jiuyi Shubang. As the documentation provides no verifiable official installation commands, directories, configuration fields, or code examples, this article only presents information verifiable on the page, and suggest referring to the official page for authoritative details. The current page displays a rating of 100000, 761 downloads, and 0 installations. As public adoption has not yet been demonstrated, it is recommended to conduct a small-scale trial run initially and perform manual verification of the facts and compliance expressions in the generated draft.
Read MoreAd Delivery Data Diagnosis | Guo Yiyi: Organizing Ad Data into Funnel Diagnoses and Testing Recommendations
**Ad Spend Data Diagnosis | Guo Yiyi** Guo Yiyi is an analytical Skill published by community author user_8d36cde0, located at @user_8d36cde0/guoyiyi-ad-spend-diagnosis. Version 1.0.0, it is categorized under data-analysis and sourced from the community. Its tag indicates `requires_api_key=false`, meaning no API Key is needed for use. Page metadata shows a rating of 100000, 818 downloads, and 0 installations. It is designed to take four types of input data—ad campaigns, creatives, conversions, and costs—and generate funnel diagnostics, issue prioritization, and testing recommendations. It targets advertisers and growth leads. The article details these three capabilities, explaining that it operates as a single-input, single-output analysis process without automated data fetching or ad execution. The installation and activation section states that as of the time of writing, the two official addresses (api.skillhub.cn and www.skillhub.cn) were inaccessible. Therefore, installation commands, the Skill directory location, and any differences between Cursor, Codex CLI, and Claude Code could not be verified. No speculative commands are included; only the official addresses are provided for future verification. Typical usage lists only the four input categories and three outputs explicitly mentioned in the brief. It notes that specific input requirements, such as field formats and time ranges, could not be verified. The applicable scenarios and caveats section highlights that diagnostic quality depends on data quality, and that priorities and testing suggestions require manual confirmation. It also advises verifying page availability and version before installation. In conclusion, the article reviews the tool's value in packaging ratio calculations and "what to fix first" decisions into a single input-output operation, and provides the official URL.
Read Moredsh-persona-kit: Swap personas, tune tone, track relationships inside DeepSeek Harness
dsh-persona-kit is a DeepSeek Harness plugin (MIT 0.1.0) maintained by xsllo425. On the host side, it provides five preset personas, a 1–5 exaggeration-level slider, per-persona relationship growth, 40 achievements and reincarnation, reply avatars, and optional gray-text input suggestions. Installation requires DSH 0.2.1-rc.x and Node ≥22.13; install via `dsh-plugin.cjs link` and run `install-examples.mjs --presets`. Enable it in `~/.dsh/persona-kit/config.json` and restart; during conversations, use `/role` to switch personas. Except for Harness model requests, no data leaves the local machine, and gray-text suggestions are disabled by default to avoid extra calls.
Read Moredsh-agent-menu：Simplifying Right-Click Menus for DeepSeek Harness
dsh-agent-menu, maintained by xsllo425, is a plugin for the DSH client that provides uncluttered context menus for conversation areas, input boxes, and the sidebar: copy selection & code, link & file operations, image copy & save-as, and supplements the workspace/session native menus with copy directory, reveal in, and session deletion (localized). It does not insert AI prompts. Install via `dsh plugin --profile desktop add github:xsllo425/dsh-agent-menu`; requires dsh >= 0.2.1-rc.2; MIT license.
Read MoreTemplate Customization: Lock the AI-Generated Draft Minutes into the Company's Own Format
Meeting Minutes Template Customization Skill leverages Tencent Meeting (TMeet) recording, transcription, and intelligent minutes data. Using a "learn → generate → adjust → consolidate" loop, the skill refines an AI-generated initial minutes into a customized template aligned with enterprise conventions. Users select one of eight preset structures as a starting point: General Meeting, Decision-Oriented Meeting, Action Tracking Meeting, Retrospective Review, Daily Stand-up, Brainstorming Session, Executive Meeting, and Government/Party Meeting. The initial version is generated from a meeting recording. Users can provide authentic historical minutes to assist customization and provide feedback in natural language to iterate the template until it meets their needs. Once satisfied, the result can be finalized as the default template for that meeting category and applied automatically to subsequent similar meetings, and can be converted to scheduled tasks to generate meeting minutes automatically as soon as meetings end. Government/party meeting scenarios use a dedicated official-document style sub-framework that covers decision-oriented meetings, administrative routine meetings, thematic meetings, and symposium/workshop meetings, following the terminology and usage standards specified in GB/T 9704-2012. The TMeet skill is required; Tencent Docs skill and scheduled task capabilities are optional.
Read Moredsh-keepalive：Auto-restart DSH Desktop on Disconnect or Crash
dsh-keepalive is a DSH admin-security plugin maintained by XnQM-DEV, Windows-only with zero npm dependencies. The watchdog runs from dataDir/runtime via scheduled tasks. It provides heartbeats every 5 minutes, auto-restart on disconnect/port no-response (clearing login cookies before restart), maintenance mode, health backup and restore for cordis.patch.yml, configurable daily backups and disk cleanup, and OneBot/webhook/command notifications. An optional host half-side writes heartbeats and detects long conversations stuck. Installation flow: init, doctor, install --elevate, optionally link-into-dsh or dsh plugin add dsh-keepalive.
Read MoreRPM-38-Academic Visit Topic Planning: Delegating Pre-visit Topic Sorting to the Agent
rpm-38-Academic Visit Topic Planning (Tuojiang) is a professional-type Skill published by the Baitong AI Service Platform and distributed via the platform's authorized packages. Version 2.0.0, coordinates @baitongai/rpm-38-academic-visit-topic-planning, and it requires an API Key. The official page provides only a one-sentence description and does not disclose the tool list, input/output formats, installation commands, or configuration fields. Therefore, this article focuses solely on verified information: source attribution, coordinates, version, the `requires_api_key` tag, as well as page metrics such as 0 installations and 775 downloads. The installation path depends on the Baitong authorization process. Before integration, users must obtain the package to review the SKILL.md file, confirm the capability boundaries and output formats of the topic planning feature, and then conduct minimum scenario validation. The article outlines the verification methods for the coordinates, version, and tags, along with a four-step validation approach, making no speculation regarding unverified features.
Read MoreZhuanzhuan Second-hand Assistant: A Skill Integrating Zhuanzhuan's Second-hand Transaction Capabilities
"Zhuanzhuan Second-hand Assistant" is published and maintained by Zhuanzhuan and belongs to the business-ops category of enterprise source Skills. It is currently at version 1.5.3 and does not require an API Key. It wraps the second-hand transaction services of the Zhuanzhuan platform into accessible capabilities, covering four types of functions: finding second-hand quality items, checking transaction market trends and price movements, photo-based valuation with door-to-door pickup for recycling, and advice on when to sell for better returns. It supports categories such as mobile phones and digital products, luxury goods like LV and Rolex, hobby items, and used cars such as Tesla, along with platform guarantees including official authentication, authenticity assurance, and worry-free after-sales service. The documentation does not provide installation commands or installation differences across Cursor / Codex CLI / Claude Code, nor does it offer reproducible call examples. For integration, it is recommended to refer to the official page instructions. Public data shows a rating of 100000, 835 downloads, and 0 installations, indicating low adoption. It's advisable to validate on a small scale before integration. It is suitable for developers who need to provide second-hand transaction capabilities in their products or Agents, but market trends and valuation results are affected by second-hand price fluctuations and actual product condition, requiring self-designed fallback logic.
Read Moredsh-browser-bridge: Driving Open Edge/Chrome in DSH
dsh-browser-bridge is a DSH client plugin maintained by xkzm123 that bridges via UI Automation, SendInput, and optional Tampermonkey user scripts to operate local, already-running instances of Edge or Chrome on Windows 10/11 without requiring a debug port or browser restart. Install it via the command dsh plugin --profile web add github:xkzm123/dsh-browser-bridge, followed by a DSH restart. The plugin alone provides about 14 tools, expandable to 22 when combined with bridge.user.js. It is ideal for Harness developers who need to read pages, click, type, and batch-process within real browser windows while preserving login sessions. Review the MIT-licensed source code before use, and note that the plugin injects input locally with the permissions of the current DSH process.
Read MoreMiddle Management Assistant: Delegating Daily Management Tasks to Reusable Skills
Skill for middle manager daily assistant scenarios, identified as `@user_9d5a2a39/mid-leader-assistant`, categorized under `business-ops`, tagged `requires_api_key=true`. Built on a local data foundation, it covers the full workflow of logs, calendar, tasks, employees, 1-on-1 meetings, OKRs, report decks, emails, performance, search, and Tencent Meeting. Version v3.0 introduces personnel management (team building/training/skill matrix/talent review 3x3 grid/turnover warning/PIP), cross-department collaboration, upward reporting templates, and deep team analytics (workload balancing/sentiment trends/collaboration network/human efficiency), along with the matplotlib charting engine and postcheck validation. Since v3.1, chart PNGs, the SQLite database `leader_assistant.db`, and Excel exports are configured to output to the tempdir by default, avoiding rejection of binary uploads. v3.2 fixes four latent bugs in the backup list, collaboration network, sentiment trends, and turnover warning, and adds employee attribution support to work_logs. v3.3 adds a pre-meeting brief generator, meeting cancel/delete tools, extended employee search, five content templates, LogManager's `employee_id` filter, and shared utilities in `scripts/utils.py`. It offers two modes: a plain-text zero-dependency mode and a full Python-featured mode (SQLite, Web dashboard, 114 MCP tools).
Read MoreYouTube Video Summarizer: From Subtitles to Structured Summary
This article introduces the SkillHub community Skill "YouTube Video Summarizer" (ID: @user_9e3b7930/youtube-summarizer-cn, Version 1.1.0, Category: knowledge-management). It downloads YouTube subtitles and generates structured summaries containing core insights, key quotes, and highlights. It defaults to Chinese output, supports multiple languages, and is compatible with all URL formats including watch, youtu.be, shorts, embed, and live. The tag is `requires_api_key=false`. The article first explains how this tool solves the problem of labor-intensive video summarization, then lists its core capabilities and trigger methods. It subsequently outlines installation prerequisites: since official page scraping may fail, it avoids listing unverified commands, providing only the source, classification, permissions, and operational requirements. Next, the article discusses applicable scenarios and limitations, including dependency on subtitles, network accessibility, language selection, and copyright considerations. Finally, it provides the official page address and notes that this community Skill currently has 807 downloads and a rating of 100000, suggesting users test it before adoption.
Read MoreGLM Quota Monitor: Check Zhipu GLM Coding Plan remaining quota in DSH.
glm-quota-monitor is a DSH community plugin that displays GLM Coding Plan's 5-hour/weekly token windows and built-in tool 30-day usage limits via the Zhipu monitoring API. It supports settings page details, composer dock, and the Agent tool glm_quota_check. By default, it reads ZAI_CODING_CN_API_KEY from the credential store, and refreshing does not consume plan quota. Installation is done by cloning and running plugin_manager install_bundle, with optional credentialRef and endpoint configuration.
Read MoreTransforming Enterprise Communication Scripts into Reusable Assets
Introduction to the "话术" (Scripts) Skill on SkillHub, Released by Kunlun Growth, v1.1.0 (business-ops category) This skill released by Kunlun Growth on the SkillHub platform consolidates communication scripts across various scenarios—including sales, negotiation, customer service, interviews, and upward communication—into proprietary assets that can be retrieved based on scenarios and adapted to specific contexts, without requiring an additional API Key. The article outlines the core capabilities, the method for enabling through SKILL.md, its typical usage, and contingency measures.
Read Moredsh-xixi-cost: Estimating RMB Costs per Turn and Conversation in the DSH Web GUI
dsh-xixi-cost is a DSH client plugin maintained by xixi0v0. It displays estimated CNY reference values based on DeepSeek's public price list in a statistics row at the bottom of each reply and below the input box in the Web GUI. It supports clickable details for cache hits/misses and output bucket breakdowns. This is a pure frontend bundle installed via `dsh plugin --profile desktop add github:xixi0v0/dsh-xixi-cost`, requiring a DSH restart after installation. Amounts are estimates, not official charges; session totals are based on the full log projection `xixiCost`. MIT licensed, version 0.1.2.
Read More