dsh-usage-dashboard: Viewing Balance, Cost Estimates, and Peak/Off-Peak Hours in the DSH Web Sidebar
Introducing dsh-usage-dashboard, a DSH client plugin maintained by MrXiongBear: displays DeepSeek balance, local token-based consumption estimates, and peak/off-peak time countdowns in the bottom-left corner of the web interface. Click to expand the full usage dashboard and switch the default LLM. Describes the installation command, configurable sidebar items, per-task statistics, and known limitations (2026 holiday schedule, pricing requires manual updates). Data is displayed locally only, MIT licensed.
Read Moredsh-task-steward: A Deterministic Process Manager for DSH Tasks
dsh-task-steward is a DSH admin-security plugin released by MrXiongBear, which uses rules and a state machine to supervise running tasks: handling stuck states, idle spinning, no progress, forbidden commands, privilege calls following taint, budget and privilege escalation requests, among others. The default policy is "report," which only writes to the ledger without intervention; switching to "auto" engages a four-level escalation of actions: record/nudge/block/stop, and does not default to allowing stop. It is not published to npm; you must link it into your profile and register it in bundles. The ledger and boot.json can self-verify whether it has been loaded. MIT license, Node >= 22, version 0.1.3.
Read MoreDeveloper Focus Flow: Session Tracking & Task Board - Command Line Work Session and Task Management
A command-line productivity tool for developers, designed to track work sessions, manage tasks, and set goals. Maintained by community author user_2fd890c9, the current version is 1.0.8. Indexed under the dev-programming category on SkillHub, it is tagged as a community source (no API key required), with a rating of 100000 and 854 downloads. This article introduces its core capabilities (session tracking, task board, goal setting), coordinates, and official address. It notes that installation commands and specific usage should be based on the official page; as the official page could not be accessed for verification during the writing of this article, no unverified commands or configurations are provided. This tool is suitable for developers who prefer terminal-based workflows. Given that it originates from a community personal account with zero install counts, it is recommended to read its SKILL.md and independently verify its stability before use.
Read Moredsh-task-intake: Clarification and optimization before new tasks
dsh-task-intake is a DSH workflow bundle maintained by MrXiongBear that injects `task-intake:protocol` into the system prompt. Before executing any new task, it restates objectives, surfaces assumptions, retrieves prior art, proposes bounded improvements and questions, and then provides an optimization plan. Based on the GitHub README, this article explains the install command `dsh plugin --profile desktop add`, the `cordis.patch.yml` configuration option, the conditions under which intake can be skipped, and the all-or-nothing profile installation risk, to help developers evaluate whether to enable it on the desktop profile.
Read MoreEnterprise Risk Radar: Transforming Risk Scoring and Sentiment Monitoring into Readable HTML Dashboards
This article introduces the community Skill "Enterprise Risk Radar" (coordinates: @user_053d27d0/risk-radar, version 1.0.0, categorized under business-ops). According to its official description, it provides six-dimensional risk scoring, public sentiment monitoring, red-yellow-green traffic light alerts, and an HTML dashboard, aiming to consolidate scattered enterprise risk signals into unified scores and readable dashboards. The documentation notes requires_api_key=false, but the captured page lacks specific installation commands, parameter examples, or red-yellow-green threshold definitions. The article outlines verifiable official entry points, applicable scenarios, and caveats including its community origin, undisclosed thresholds, and unverified sentiment coverage.
Read Moredsh-settings-hub: Centralizing third-party plugin settings under "Plugin Settings"
dsh-settings-hub is a DSH client plugin maintained by MrXiongBear that consolidates third-party settings.sections—excluding built-in sections such as account, general, and model—into the "Plugin Settings" page. It supports search and in-page expansion, reusing the original components and inject. Based on the repository README, this article introduces the collapse rules, mirroring mechanism, plugin_manager installation method, profile backup, and experimental risk warnings.
Read MoreCulture and Tourism WeChat Official Account Information Source: Filtering Viral Articles by View Count and Generating HTML Reports
The Cultural Tourism WeChat Feed is a general-purpose Skill in SKILL.md format sourced from the community, with the identifier @user_5f9c21aa/cultural-tourism-wechat-feed-redfox, version 1.0.1, and categorized under knowledge-management. Addressing the pain points in cultural tourism content operations where "WeChat Official Accounts are scattered and hard to find, and article popularity is difficult to assess," it automatically searches for popular articles from cultural tourism-related Official Accounts, filters the most viral content based on read counts, and generates a formatted HTML report. Its trigger words include 文旅公众号, 文旅信息源, 文旅资讯, 旅游公众号, 景区文章, and 文旅爆款. The article introduces its positioning and capabilities, official URL, the configuration prerequisite of an API Key, and reproducible invocation examples. It notes that installation commands were omitted due to missing information in the source materials. It also outlines the target audience and limitations—such as covering only WeChat Official Accounts, inconsistent metrics for read counts, and the need for a trial run given it is community-sourced with zero installs. Finally, it provides the official URL.
Read Moredsh-orchestrator-canvas: The Five Orchestration Gates and the Read-Only Canvas of "Truth = Graph"
dsh-orchestrator-canvas is a DSH workflow plugin maintained by MrXiongBear that implements the Brief→Decompose→Dispatch→Verify→Accept pipeline as an orch_* toolchain and run.json evidence chain, composing host subagents rather than writing its own scheduler; it provides a read-only "Orchestration Canvas" overlay for the input bar. Installation requires linking into profile bundles and restarting after auditing, compatible with DSH 0.2.1-rc segment and desktop profile.
Read MorePharmaceutical Industry HR Organization: A Skill for HR Perspective Analysis of Pharmaceutical Sales Teams
Healthcare Industry HR Organization Skill, targeting pharmaceutical sales and marketing teams, provides an analytical framework across eight core dimensions from the HR and marketing organization perspective: organizational design, headcount planning, talent standards, compensation and performance, career levels, training and development, workforce efficiency diagnosis, and compliance red lines. Published by SkillHub user user_31815b66, community-sourced, professional category, version 1.0.0, no API Key required. The Skill embeds the foundation.md base module, dynamically adjusting output depth based on junior or senior tiers, and strictly distinguishes three functional categories: HR-exclusive functions, collaborative interfaces, and non-HR boundary functions. Designed for scenarios requiring organizational diagnosis of pharmaceutical marketing teams, headcount estimation, compensation and performance design, career level system construction, recruitment and selection, workforce efficiency analysis, or compliance red line mapping.
Read MoreConsumer Rights Protection Toolkit: Organizing Consumer Protection Law and Redress Mechanisms into a Callable Knowledge Base
Introduction to the Consumer Rights Protection Toolbox on SkillHub (handle: @frommax/consumer-rights-toolbox-v3; version v1.1.0; released by Fumaq Craft Brand Strategy Consulting Agency; listed under the professional category; paid, no API key required). It internalizes the key points from 15 monographs covering consumer rights, product quality, food safety, online transactions, e-commerce law, automotive rights protection, and fraud prevention for older adults, and layers in 2024 updates such as the Regulations on the Implementation of the Law on the Protection of Consumer Rights and Interests and the Supreme People’s Court judicial interpretation on the trial of online consumer dispute cases. It covers consumer dispute diagnosis, the nine consumer rights, five dispute-resolution pathways, punitive damages, as well as evidence checklists and document templates for six scenarios: online shopping, livestream sales, prepaid cards, automobiles, food safety, and fraud prevention for older adults. The article explains its knowledge base, five functional layers, and three usage modes, along with three limitations: it does not provide legal representation; currency depends on timely maintenance by the provider; and monographs are distilled to key points and must be verified against the originals. Because the official page could not be scraped, the installation command was not included; only the listing coordinates, payment status, and the no-API-key requirement were confirmed.
Read Moredsh-capability-ledger: DSH Profile Capability Ledger and Security Check
An introduction to the DSH plugin `dsh-capability-ledger`, maintained by MrXiongBear. It performs read-only aggregation of skill and tool usage from persistent session logs, tracks fixed per-request tool-schema costs, and conducts security audits on installed bundles. It clarifies that while the current dashboard can be removed from the desktop profile, the `audit-cli` and `preflight-restart` remain functional. Additionally, it covers link installation, the "restart gate" triage, statistical caliber, and privacy boundaries.
Read MoreCyber-self: Build a Local Repository for Every Identity
cyber-self (Cyber Mirror) is a DSH memory plugin maintained by MrLsss: each identity corresponds to a workspace and a local SQLite database, with support for multi-format import, automatic chunking, full-text and Ollama bge-m3 vector indexing, and conversational Q&A with [1][2] citations. It can be installed from the DSH Desktop plugin marketplace, or loaded into a custom profile via `dsh plugin --profile kb add github:MrLsss/cyber-self`; data is stored in `.cyberself/kb.db`, and identities do not perform cross-database retrieval.
Read MoreQBS Problem Solving Skill: Locate Construction Acceptance Code Clauses by Keyword or Standard Number
# Introduction When engaged in development or consulting related to municipal engineering, construction, and water utilities, the most common action is checking standards and codes. Typically, knowing a keyword (e.g., "foundation pit monitoring" or "water tightness test") or a standard number (e.g., GB 50268), one needs to pinpoint the specific manual, chapter, and clause to cite them in code, documentation, or proposals. The challenges of maintaining a private standards library are direct: standards undergo revisions, clause numbers change, inclusion scopes are difficult to define, and version inconsistencies easily arise across multiple teams. The **QBS problem-solving Skill** turns this step into a callable function: given a keyword or standard number, it returns the corresponding construction and acceptance codes and locates the specific chapter and clause. # What is this The **QBS problem-solving Skill** is a Skill designed for standards retrieval scenarios, with the coordinate `@
Read More@moyumole/dsh-plugin-updates: Checks against npm and updates third-party DSH plugins in the Profile in place
The DSH plugin `@moyumole/dsh-plugin-updates` (Moyumole, v0.3.3) provides a read-only summary of third-party bundles within the Profile via a Guide card on the right, a "Plugin updates" tab on the Plugins page, and a badge on the left Plugins icon. It queries npm for the `latest` version and uses the official `pluginManager` to perform in-place updates via `removeBundle` and `installBundle`. Features include re-querying every 30 minutes while the page is visible, a 10-minute registry cache, "Update all" (skipping packages with a decompressed size over 30MB), and rollback on failure. The Host requires no additional permissions, and network access is limited to read-only npm queries. Verified on DSH 0.2.1-rc.3 for Windows. Installation example: `dsh plugin --profile web add "@moyumole/dsh-plugin-updates"`.
Read MoreWeChat Publishing Pipeline: A 9-Step Workflow from Prompt to Draft
The "WeChat Official Account Writing & Revision Pipeline" is an Agent Skill built on the universal SKILL.md format, identified as @user_bf4c5610/gzh-article-pipeline (Version 1.0.0). Sourced from the community and categorized under content-creation, it requires no API Key. Taking multiple topics provided by the user as input, it automates the entire workflow through a 9-step process: topic selection → information gathering → framework drafting → content enrichment → AI-touch removal → image selection → layout adjustment → cover image + summary → draft publication. After completing information gathering in Step 1, it recommends 3-5 topics for user confirmation. The skill generates two versions of cover images: 2.35:1 wide-format and 1:1 square, and keeps the summary within 120 characters. It enforces strict constraints such as not directly copying reference materials, not fabricating data, leaving blanks for missing data pending confirmation, not misusing original images, and offering placeholders and prompt suggestions for images requiring regeneration. Ideal for individual WeChat Official Account authors and content operators, the skill ultimately generates a draft in the backend instead of automatic mass distribution.
Read Moredsh-preflight: DSH Plugin Pre-installation Check, Post-installation Inventory, and Pre-update Assessment
dsh-preflight is a DSH management plugin maintained by mouyu6508-hue, offering three workflows: `preflight_check`, `preflight_inventory`, and `preflight_updates`. It performs pre-installation validation of `dsh.bundle`, engine/peer, native compilation, and loader name conflicts; conducts post-installation inventory of installed plugins grouped by capabilities; and compares manifests before updates to report only newly introduced risks. Supporting both a standalone CLI (including `--json` output) and an in-session health check panel, it fetches data via read-only APIs. Licensed under MIT, it requires Node.js 18+ and a DSH environment that supports bundles. Please refer to the SkillHub directory page and the GitHub README for installation commands.
Read Moredsh-fusion: Conflict Detection and Configuration Fusion in DSH Plugin Assembly
dsh-fusion is designed for DeepSeek Harness layered profiles to detect issues such as cross-bundle duplicate loaders (install-shadow) and configuration keys being silently overwritten by subsequent layers. It provides an independent CLI (usable when DSH fails to start) and a fusion_report tool, supports dual-path analysis via static patching and runtime loaders, and can generate cordis.fusion.yml. Maintainer: mouyu6508-hue. MIT License. Node 18+.
Read Moredsh-memory: DSH Local Memory and Compression Pipeline
**dsh-memory** is a memory-type plugin for DeepSeek Harness maintained by mounta1n-18. It has zero dependencies and requires no build step, with data stored in `~/.dsh/memory/`. It provides a one-line classified outline in the system prompt, four tools—`memory_read`/`write`/`update`/`delete`—a three-part lesson structure, and memory management on the settings page, together with recall statistics and a compaction pipeline triggered every 100 turns or by the hourglass. Sessions also feature a memory log page and `/remember` conversation summarization. Installation is only supported by specifying the absolute repository path via `plugin_manager install_bundle`; Host changes require restarting DSH.
Read Moreboss-zhipin-assistant: BOSS Zhipin Recommendation Page Automated Application Skill
This article introduces the "boss-zhipin-assistant" Agent Skill. Operating within a logged-in Chrome browser, it leverages Kimi WebBridge (Port 10086) to control the recommendation page. It first scrolls to trigger lazy loading for all job listings, then utilizes an Agent to perform semantic decision-making and apply to jobs individually. The system utilizes a three-layer loop architecture consisting of keyword evolution and serial application on the recommendation page. The skill features environment self-diagnostics, resumable application tasks, and daily goal tracking, generating an HTML report with WeChat and Alipay tip QR codes upon completion. The official documentation notes that version v5.5 replaces decision.json with a shell_executor and Python overwrites to eliminate confirmation popups within the loop. The article outlines the operational workflow, prerequisites (Port 10086, Chrome login status), applicable scenarios, and limitations, while noting that official documentation does not provide installation commands. Published by user_f5c8032a from the community, this SkillHub entry is categorized under business-ops, with version 26.8.5, a score of 100,000, 929 downloads, and the tag requires_api_key=false.
Read MoreScript Writing: Batch Synthesis of Unique Topics and Generation of Voice-over Scripts
Introduce the SkillHub community Skill 'Copywriting' (@user_07f45fa7/jaie03, version 1.0.0, content-creation). From a blogger analysis library's topic pool, hooks pool, style pool, and copy templates, it synthesizes unique topics, calls a real LLM to generate spoken-only scripts, and outputs a DOCX containing the topic, pain points, reference hooks, spoken-only body text, and a human-sounding score. The default campaign brand is Huawei HarmonyOS Smart Home, with the deny-list switching accordingly; you can register other brands or industries in references/brand_policy.md. Use cases include writing copy, writing a spoken script, generating viral copy, producing several scripts, and finding topics, etc. It does not depend on a single video; single-video imitation should use video-deconstruct-pro (VDP). Tag requires_api_key=true; page metrics: 100000 ratings, 902 downloads, 0 installations. The scraped materials do not provide the full SKILL.md or an installation command; the body text does not fabricate an installation script.
Read MorePresentation on Specialties, Courses, and Specialty Clusters in Higher Vocational Education & Smart Assistant for New Program Applications: Integrating Industrial Logic into Teaching Presentations
An Agent Skill designed for vocational college teachers, curriculum leaders, program heads, and teaching research section directors. Located at @user_63830f0e/gaozhi-shuo-kecheng (Version 1.6.1), this community-sourced tool is categorized under education. It targets four scenarios: course presentations, program presentations, professional group presentations, and new program applications. It generates readable verbatim scripts, PPT outlines, and predictive expert Q&A strategies. The writing follows a主线 of industrial changes → job transformation → typical work tasks → vocational competencies → curriculum modules, analyzing through three perspectives: industry, competency, and AI. It aligns with key teaching reform elements, the New "Double High" Plan, 2025 new teaching standards, and AI-empowered education initiatives. Built-in features include expert psychological models, six disciplinary lenses, eight context adaptations, anti-AI-tone writing rules, and self-check scripts. It supports four duration options (10/15/20/30 minutes) and five scenario types (secondary vocational, higher vocational, vocational undergraduate, applied undergraduate, and public basic courses). No API Key is required. Note: Industrial data and specific policy details must still be verified and filled in by the user according to their specific school and provincial regulations.
Read MoreDSH-Time-Memory: A DSH Chat Memory Plugin with Time-based Archiving and Tiered Recall
dsh-time-memory provides DeepSeek Harness with time-based JSONL writing and hierarchical expanded retrieval through memory_recall; if no match is found after two expansions, the only permitted response is "I forgot". Based on the official README, this article covers installation (entering the absolute path in the GUI after cloning), memory_save/memory_recall, configuration options, and token characteristics, and explains known limitations such as keyword-based rather than vector retrieval and profile-level personas.
Read MoreFunctional Requirements Breakdown Tool: Turning Ambiguous Requirements into Actionable Implementation Plans
The Feature Planning Tool (@user_7871dce1/feature-planning-zh) targets four scenarios: new feature development planning, complex requirement analysis, iterative task breakdown, and technical solution evaluation. Taking a single natural language description of a functional requirement as input, it outputs a detailed implementation plan featuring clear task decompositions and technical approaches. It addresses planning-phase questions such as "How should this requirement be broken down?", "Help me split this feature into development tasks," and "How to execute complex requirements?". Specifically, it answers what blocks a feature should be divided into, what each block does, and how to implement it technically, rather than detailing how to build the feature itself. Published by community author user_7871dce1, this Skill is categorized under business-ops, sourced from the community, and based on the standard SKILL.md format. It is version 1.0.0 and does not require an additional API key (requires_api_key=false). When using it, sufficiently specific requirement descriptions must be provided; otherwise, the breakdown may remain at the framework level. As its output is a planning scheme, it cannot replace coding implementation or architecture review. The article also outlines installation and activation channels (the SkillHub page and repository API endpoints), typical usage prompts, and the target audience. Given the release data—version 1.0.0 with 867 downloads and 0 installations—which warrants attention, it is recommended to test it with your own requirements before deciding whether to adopt it.
Read MoreM&A Review Key Points Assistant: A-Share M&A and Restructuring Audit Research Skills
This article introduces the SkillHub community Skill "M&A Review Key Points Assistant" (v2.0.8, professional, no API Key required). This Skill is based on the IMA M&A and Restructuring Practice Case Database, covering seven categories of A-share M&A and restructuring scenarios: major asset restructuring, share issuance for asset acquisition, restructuring for listing, cash acquisition, related-party transactions, control rights transactions, and supporting financing. It enters the corresponding context through trigger words such as merger & acquisition, M&A review, M&A restructuring review focus, transaction characterization, and M&A inquiry. The article explains its positioning, core functions, activation method based on SKILL.md, typical usage (marked as inferred since the official page is inaccessible), applicable scenarios and limitations, and provides the official address for verification.
Read Moredsh-session-groups: Freely Group Sessions Across Workspaces in the DSH Sidebar
dsh-session-groups is a DSH client plugin maintained by momijijin that adds a free grouping view on top of the official workspace sidebar: it allows sessions from different workspaces to be grouped together, modifying only the plugin's local records without altering official workspace ownership or logs. It supports nested groups, drag-and-drop, search, archive filtering, and local backup/restore; an optional session_groups_control enables AI-assisted group planning. Installation is recommended by searching for dsh-session-groups via dshmarket, and requires DSH 0.2.x with a verified Stable/official desktop shell.
Read More