Community Private Domain Operations Master: Six-Step SOP Skills for Full-Link Community Operations
Community Private Domain Operations Strategist is a comprehensive skill tailored for end-to-end community private domain operations. Published by user_942679f1 under the "business-ops" category (v1.0.0), it standardizes private domain operations into a six-step SOP: Acquire, Nurture, Activate, Explode, Rescue, and Diagnose. The skill is equipped with a "Five-Pillar" private domain capability model, a comprehensive scenario-based script library, conversion and user-fission mechanics, anti-suspension strategies, and strict compliance red lines. Special adjustments have also been tailored for silver-hair (elderly care) scenarios and temporary groups for government affairs activities. The Skill provides 11 quick shortcuts covering group matrix construction, welcome messages, new member onboarding, daily scheduling, flash groups, re-engaging dormant users, Moments copywriting, crisis firefighting, user stratification tagging, fission campaigns, and community health diagnostics. It seamlessly integrates with event planning, official account operations, short video scripting, and post-campaign review skills, feeding conversion-layer data directly into the wrap-up and review process. It is ideal for teams or individuals aiming to build a private domain from scratch who require standardized community operation workflows and bulk generation of operational scripts.
Read MoreMT Photos: Encapsulating the NAS Photo Album API into a Skill with 7 Capabilities
MT Photos is a skill from the SkillHub community (ID: @user_4df20d12/mt-photos), Version 1.0.0, published by user_4df20d12 and categorized under "ai-agent." It encapsulates the NAS album API into seven core capabilities: admin-side user, folder, and file management, as well as non-admin-side album management, image/video recognition, reading, analysis, and batch operations. All deletion operations include built-in secondary confirmation guards. The skill is tagged with `requires_api_key=true`, indicating that an API key is required for invocation. This article covers its positioning, verified capabilities, prerequisites for activation, and applicability boundaries. As official documentation does not provide installation commands or configuration fields, they are not expanded upon here; users are advised to refer to official sources for details.
Read Moredsh-device-link: One-click Access to Local DSH from Mobile and Remote Devices
DSH Desktop binds the 127.0.0.1 address and cannot be accessed directly from external devices. The community plug-in dsh-device-link (v1.0.0, Windows 10/11) presents a Wi-Fi icon and three toggles in the sidebar to expose a LAN entry point (default port 8900) and an EasyTier networking entry point (port 8899) through a session-injecting local gateway. Toggling the switches off immediately revokes access. The plug-in does not alter any DSH or EasyTier configurations. Installation via plugin_manager install_bundle. Read the README security warnings: connecting to any exposed entry point grants full DSH privileges on the host. MIT licensed; maintained by wwj-wwj1.
Read MoreWorkplace Emotion Management: A "Emotion + Balance + Decompression + Rejection" Dimensional Framework Skill for Professionals
Workplace Emotion Management is a comprehensive multi-dimensional framework Skill for workplace scenarios, covering "emotion management + work-life balance + stress reduction + anti-involution + reasonable rejection." With the identifier @user_aad6add3/workplace-emotion-manager, it is community-sourced, categorized under life-service, and currently at version 1.0.0. Its tag requires_api_key=false indicates that it can be enabled without needing any extra API key. By breaking down workplace stress into 5 major categories and 20 sub-dimensions—covering boundaries, workload and anti-involution, emotion management, work-life balance, managing up and rejection, and long-term growth—it equips Agents with a unified and complete framework for offering advice, eliminating the need for generic platitudes. This article introduces its positioning, core functions, installation and activation methods based on the standard SKILL.md, typical use cases, applicable scenarios, limitations, and the official URL for reference
Read Moredsh-chain-check: Locate the DSH to QQ link breakpoint with a single command
dsh-chain-check (wuying1630) is a DSH client diagnostic plugin that utilizes six read-only checks to identify connection breaks across DSH, surfcore bridging, and OneBot/SnowLuma to QQ. It supports direct execution via `node check.mjs` or registration of the `chain_check` tool via `dsh plugin add`, requiring Node 18+ with zero dependencies. The optional `--fix` flag automatically aligns `dsh.baseUrl` or launches the bridge; NapCat/AstrBot can bypass bridging or OneBot via `chain.config.json`. Licensed under MIT and compatible with DSH 0.2.x.
Read MoreH5 Mini Game Generator: Compliant Generation of H5 Mini Games Based on Package Specifications
H5 Mini Game Generator is an Agent Skill published on the SkillHub community based on the universal SKILL.md format. Version 2.0.0, located at @user_ce7810c4/h5-mini-game-generator and categorized as dev-programming, it does not require an API key. It generates standard-compliant lightweight HTML5 game code that adheres to H5 mini-game package specifications, keeping the main package size under 4MB. It covers gameplay genres including Snake, Tetris, Breakout, 2048, Minesweeper, Tic-Tac-Toe, Air Combat, shooting, tower defense, parkour, side-scrolling fighting, and match-3. Furthermore, it involves engineering capabilities such as Canvas gaming, object pooling, particle systems, game AI, behavior trees, wave generation, and package size optimization. This article introduces what it is, verified core features, available information versus installation sections that cannot be verified, a usage approach based on its positioning, applicable scenarios and limitations, and provides the official address.
Read MoreDSH-sidebar-shelf: Collapse DSH Sidebar Panels into a Single-Line Searchable Shelf
Each panel plugin added to the DSH sidebar occupies another row, squeezing the workspace list. dsh-sidebar-shelf, maintained by wumouren114 and licensed under MIT, is a client-side plugin that collapses all panel entries into a single “Panels” row between “New Session” and the workspace list. Clicking it expands a two-column rack in place and supports search and keyboard operations. It opens panels via the official interfaces sidebar.panellist and ctx.layout.selectPanel, and is compatible with the data-dsh-*-entry injection rows. Installation: dsh plugin --profile desktop add github:wumouren114/dsh-sidebar-shelf then reload the page to enable. Be aware that it hides the “More” menu row, and do not combine it with similar sidebar-organization plugins.
Read Moredsh-identity-relay: Persist Personality and Long-Term Memory to Disk for Automatic Loading in New Sessions
The DSH memory plugin dsh-identity-relay (Persona Relay) from wumingzhiren-lab stores persona.md, handover.md, and facts.jsonl in ~/.dsh/identity-memory, automatically injecting them at the start of new sessions and writing handover files when warned that the context is nearly full. Memories carry a five-level source rating, stamped invalidation, and same-key replacement, and the plugin provides eight tools including relay_load, relay_remember, and relay_handover, without automatically collecting chat history. Install: dsh plugin add github:wumingzhiren-lab/dsh-identity-relay#main. MIT licensed, suitable for Harness agents that need cross-session identity continuity.
Read MoreOffice Raccoon Data Comic Skill: Rewrite Knowledge Content into Character-Consistent Comics
The Office Raccoon Knowledge Comic skill (ID: @user_cae83fba/office-raccoon-comic-sl, v2.0.0, community, content-creation) transforms knowledge topics, tutorials, character stories, or product insights into comic storyboards, scripts, and image generation prompts featuring the "Office Raccoon" IP as the core character. Images are generated and compiled into a final book only after user confirmation. Key constraints include: using R-Lock asset anchors to ensure character consistency; post-processing to overlay critical Chinese text; employing a Gate-0 veto system combined with a 100-point scoring scale for quality assurance; and strictly prohibiting unauthorized image generation. This skill is not suitable for pure article polishing, poster design, PPT generation, or single illustrations.
Read MoreAI Auto Knowledge Workbench Pro: Metadata-Only Indexing and Offline Cockpit for Local Markdown / Obsidian Workspaces
AI Auto Knowledge Workbench Pro is released by Jiuqi Consulting, located at @org-qk2huurg/manage-ai-knowledge-workbench-pro. Current version 1.0.2; category: knowledge-management; source: enterprise; pricing: paid; no API key required. It targets authorized local Markdown or Obsidian workspaces. It first runs a free pre-check, and upon user confirmation sends a structurally aggregated list that contains no paths, names, or full text. After obtaining a one-time paid dedicated plan, it locally generates a metadata-only knowledge index and an offline HTML dashboard, and supports subsequent local incremental updates. This article introduces its positioning, verified core capabilities, available location and configuration details, the steps to follow the official workflow, as well as applicable scenarios and limitations.
Read MoreTargeted WeChat Official Account Harvesting · Automated IMA Knowledge Base Import: Periodically Organizing Articles from Specified Accounts into the IMA Conduit
This article introduces the SkillHub community skill "WeChat Official Account Targeted Collection · Automatic IMA Knowledge Base Import" (@user_e63d2746/wechat-mp-collect-to-ima, version 1.0.0, category: data-analysis). It subscribes to selected WeChat official accounts via wewe-rss, periodically fetches new articles, writes them into a local SQLite database, filters them by time window and title keywords, performs four-category classification, and then batch-imports them into the IMA knowledge base by category. The official description claims support for any industry, with a fully automated 9-step workflow. The tag is requires_api_key=false. Snapshot data: score 100000, 899 downloads, 0 installs. The article outlines the core features, the 9-step execution flow, trigger keywords, applicable scenarios, and caveats; the official documentation does not provide installation commands, configuration options, or reproducible examples, and the article accordingly notes that SKILL.md should be treated as the authoritative source.
Read Morefeiyu-memory: DSH Local Pure-File Long-Term Memory
feiyu-memory is an MIT memory plugin (v0.2.0) provided by wudaoyu65-collab for DSH. It structures memory as a hot layer (hot.md), an event layer (events.jsonl), a facts layer (facts.json), and per‑day summaries (summary layer). Each step performs injection according to injectTokenBudget; at the end of a turn, local rules extract information. It depends solely on Node’s built‑in modules and by default does not call remote embedding services. Installation is via bundle or `node tools/install-profile.mjs`, and the data is stored by default at `~/.dsh/feiyu-memory/`. It provides four tools—feiyu_remember, feiyu_recall, feiyu_forget, and feiyu_stats—and a side‑panel force‑directed view via HTTP on 127.0.0.1. It is suitable for agent scenarios requiring auditable, portable memory with data‑locality (no data leaving the local machine).
Read Moreempirical-analysis-stata-skill: Connecting the Stata Empirical Analysis Workflow
This article introduces empirical-analysis-stata-skill, a community Skill covering the full workflow of Stata empirical analysis. It was published by user_3611a586 on SkillHub, is categorized as professional, is version 1.0.0, has the coordinates @user_3611a586/2453678, and requires no API Key. The article first explains the Skill's positioning: it covers topic ideation, research design, data cleaning and acquisition, theoretical model development and derivation, Stata regression analysis, mechanism tests, and robustness checks, but does not address other sections of a research paper. It then lists verified publisher information, including rating, downloads, installations, and tags. The core functionality section breaks the workflow into stages and mentions methods referenced in the documentation, including CSDID, DML, synthetic DID, and causal forests. The installation and activation section only records the official page, interface endpoint, coordinates, category, source, and API Key requirements, and does not add unverified commands. Typical usage is organized according to officially stated trigger scenarios and provides guidance on structuring inputs for panel-data policy evaluation. Finally, the article outlines applicable scenarios and limitations, provides the official URL, and accurately notes discrepancies between the task description and the retrieved materials regarding ratings and download counts.
Read Moredsh-plugin-ollama-bridge: Using Local Ollama as a Batch Long Text Workstation
The `dsh-plugin-ollama-bridge` (v0.2.2, MIT license) registers four tool categories for DeepSeek Harness: `ollama_models`, `ollama_ask`, `ollama_batch`, and `ollama_lookup`. It processes batch long-form text locally on Ollama using strategies such as chunking, concurrency, deterministic format validation, and sampling labels, ensuring that the original long text does not enter the main model context window. To install, add `github:wth373737/dsh-plugin-ollama-bridge` or a local pnpm path within Plugins, and configure parameters including `baseURL`, `model`, and `numCtx`. This plugin is suitable for offloading local inference to reduce the token load on the main model. Users must distinguish between format validity and semantic correctness, and should pay close attention to permission boundaries regarding read/write operations and network access approval.
Read Moredsh-plugin-mail-triage: IMAP Mail Triage and Safe Cleanup
dsh-plugin-mail-triage is a DSH workflow plugin released by wth373737: IMAP read-only incremental scanning, hard protection rule layer and optional Ollama classification, producing green/yellow/red triage reports and hash-embedded manifests. Only after approval cards are processed does it MOVE to Trash. Credentials can be imported from Thunderbird by local decryption into the Host store; the reading and writing tools are separated and no expunge is performed. It is installed via the Plugins panel by adding github:wth373737/dsh-plugin-mail-triage.
Read MoreZifeifei Boutique Customer Acquisition: Content Generation Skills for Local Marketing in Women's Multi-Label Stores
"Z非霏 Clothing Store Customer Acquisition" is a content generation skill created by SkillHub community developer user_ae2798a1. It is Version 1.0.0, classified under content-creation, sourced from the community, and requires no API Key. Designed for multi-category women's clothing stores (apparel, accessories, bags, jewelry), it covers five types of content: Douyin/WeChat Channels scripts, Xiaohongshu (Little Red Book) promotional copy, WeChat Moments captions, WeCom/private chat scripts, and model outfit descriptions. The skill is activated via trigger words such as clothing store, women's clothing store, customer acquisition, copywriting, scripts, WeChat Moments, Xiaohongshu, Douyin, WeChat Channels, private messaging scripts, and outfit changes. The article outlines its positioning, core features, installation coordinates, typical usage, applicable scenarios, and precautions.
Read Moredsh-word-master: Using Agent Presets to Turn Materials into Standardized Word Documents
dsh-word-master is a workflow-type DeepSeek Harness plugin that provides the word_* toolchain via the Word Master preset: source material to Markdown conversion, specification freeze, section-by-section writing, quality gates, HTML review, and rendering/export to DOCX/PDF. Writing preferences are persisted to disk and injected at each round. Its architecture is modeled on PPT Master. Install via npm as dsh-word-master; requires Node >=22 and Python 3.10+. MIT licensed; maintainer wszw142857-beep.
Read Moredsh-qwen-asr: Speech-to-Text in DSH using Qwen DashScope
dsh-qwen-asr is a DeepSeek Harness plugin maintained by wsdfgf that leverages Alibaba Cloud's DashScope to facilitate automatic audio transcription and provides a `qwen_transcribe_audio` tool. It defaults to the asynchronous `qwen3-asr-flash-filetrans` model (utilizing temporary uploads and task polling, with a maximum file size of 1 GB) and can be switched to sync mode for processing short audio clips. Configuration supports either the `DASHSCOPE_API_KEY` environment variable or the plugin settings page. The tool returns recognized text and offers the option to retain the original JSON response. Installation involves downloading `dsh-qwen-asr-1.2.0.tgz` from the Releases section and adding it via the DSH plugin page.
Read MoreInvestment and Financing Skill: Streamlining the Material Processing from Identification to Validation in a Pipeline
This article introduces the investment and financing skill on SkillHub (Coordinates: @user_2fd890c9/skill-54580, Version 1.0.1). Classified under 'business-ops' with a 'community' source, it facilitates investment and financing tasks including identification, organization, generation, and validation, outputting ready-to-use result files (Tag: requires_api_key=false). Due to failed official page scraping, this article lacks installation commands, configuration fields, or reproducible examples; therefore, only verified capabilities and metadata are presented, and speculative commands are omitted. The article provides the official and page addresses, noting that the skill is community-sourced with zero installations, and there are discrepancies between the provided data and leads regarding ratings and download counts. Verification is recommended before referencing.
Read MoreWeChat Insight Officer: Deconstructing Viral Articles into Reusable Logic
"WeChat Official Account Intelligence Officer" is a content creation Skill published by Lianqi Jinfu on SkillHub (Version 1.0.1). It is a paid skill that does not require an API Key. It addresses the pain points of Official Account operators who "cannot understand why viral hits become popular" and whose "teams only copy titles when replicating success." By pasting the body text of a WeChat article, the tool analyzes it across six dimensions: title hook, structure and rhythm, emotional curve, transmission points, interaction design, and conversion support. Every judgment is supported by citations from the original text, and it outputs mimicking writing guidelines and topic selections based on similar mechanisms. The article introduces its positioning, core capabilities, official page and interface addresses, and typical use cases, as well as notes regarding its paid nature, the limitation of analysis to textual logic, and the requirement for manual rewriting when mimicking.
Read Moredsh-surface: Rendering Live HTML Pages with Callbacks in DSH Conversations
dsh-surface is a DSH client plugin maintained by woolsen that displays self-contained HTML pages via the render tool within DeepSeekRuntime Web UI conversations, with optional blocking to wait for user submission. The page calls dshSurface.submit to write JSON and summaries back to the same tool call. Based on the official README, this article introduces installation commands, tool fields, the page API, the sandbox security model, compatibility, and known limitations, and explains applicable scenarios and the necessity of reviewing the source code before installation.
Read MoreGigOn: A C-End Job-Seeking Skill for Gig Workers
YouHuo · Job Seeking Assistant is a consumer-side gig-seeking Agent Skill released by Shenzhen YouHuo Technology Co., Ltd., with the identifier @org-ov217u2x/youhuo-c-job-seeker, version 1.0.3, and category business-ops. It is activated by trigger phrases such as "find a job", "look for work", "find a part-time job", "find gig work", or "are there jobs nearby". It relies on the MCP service "youhuo-c-api" to fetch job data, and if not installed, the Agent must first guide the user through the installation. This article introduces its trigger logic, capability boundaries, activation methods, typical use cases, and usage considerations.
Read MoreThree Major Graph Construction Skill: Generating and Linking Knowledge, Ability, and Problem Graphs for Course Chapters or Units
This article introduces the "Three Major Graph Construction (Knowledge Graph, Problem Graph, Capability Graph)" skill. Targeting a specific chapter or teaching unit within a university course rather than a whole textbook, it generates three graphs—Knowledge, Capability, and Problem—along with a fusion and linkage page. It covers the definitions of these three graphs and the OBE+Bloom alignment method, introduces a workflow of "Excel confirmation followed by HTML delivery," and details a unified ID coding scheme for the entire course (K/C/P + Chapter ID + Sequence Number, with codes reserved for the full course and construction units defined as chapters/units). Furthermore, it describes a single-file interactive HTML featuring animations, cross-graph navigation, and associated questions, while sharing best practices and pitfall-avoidance experiences regarding coding and rendering distilled from practical applications. Finally, the article outlines applicable scenarios, limitations, and provides the official access link.
Read Moredsh-team-autopilot: Deterministic Automated Dispatch for Native Agent Teams
dsh-team-autopilot is a DSH workflow plugin released by wooden-tan. Mounted on `ctx.agentTeams`, it provides deterministic scheduling for native Agent Teams: it uses `sendMessage` to wake members when tasks are ready, reclaims tasks upon stalls or failures, and prevents overlapping dispatch via soft locks based on `writeScopes`. It does not create a secondary team storage layer or inject policy text into the main agent. To install, you must first overlay `@deepseek-ai/dsh-experimental-agent-team-profile`, then add the plugin using npm or `dsh plugin --profile web add`. Maintenance for the project ceased on 2026-11-02, with the final version being 0.6.7 under the MIT license.
Read MoreSenior Accountant: Skills for Chinese Accounting Standards
senior-accountant is a professional-category Agent Skill released by the community, version 1.0.2, located at @user_8f4be772/senior-accountant. It integrates CFO-level financial decision-making, auditing by signing CPAs, tax agent tax-related services, CMA decision support, and senior bookkeeping into a fixed workflow, and is scoped within the three frameworks of Chinese Accounting Standards (CAS), the Chinese tax law system, and Chinese Auditing Standards. Its trigger conditions cover financial statement preparation and review, audit engagements, tax planning and annual tax settlement, managerial accounting, investment and financing decisions, journal entries, compliance reviews, CPA/CTA/CMA exam preparation consultation, and interpretation of fiscal and tax policies. There is no command-line installation method in the materials, and only requires_api_key=false can be confirmed. It provides judgments and processes without directly connecting to tax authorities or ERP systems.
Read More