dsh-token-dashboard: Aggregating DSH Local Sessions into a Token Usage Dashboard
@shi-yangyang/dsh-token-dashboard is a Cordis client plugin on DeepSeek Harness. Through `sessionPersistence`, it reads local session logs in read-only mode, aggregates Token, request, and cache metrics for all sessions under the current profile using the same four-bucket accounting method as `dsh-token-meter`, and provides overview, model, session, request log, and cost estimation pages within the DSH Slot. It supports linkage of time ranges, scan intervals, and instant synchronization, and includes built-in price snapshots for OpenRouter as well as DeepSeek peak pricing. Costs are strictly local estimates. Installation is via a locally built bundle (`plugin_manager install_bundle` or the Plugins interface), requiring Node 22.18+ and pnpm, and has been verified on DSH 0.2.1-rc.3.
Read MoreMSL Visitation Scripts: Consolidating Academic Visit Phrasing into a Reusable Framework
MSL Visit Script (@baitongai/msl-visit-script, v4.2.0, BaiTong AI Service Platform) is a Skill designed to generate MSL academic visit scripts based on the STEPS framework. From the perspective of a senior Medical Science Liaison, it integrates the five stages of the physician cognitive cycle, the six-step visit approach, LSCPA objection handling, FABE value transmission, GRADE evidence grading, Milton covert language patterns (MSL-specific low-dose version), the DA linkage mechanism, and Compliance Circuit Breaker 2.0, to output a library of academic visit scripts covering the full lifecycle of physician cognition. Its core differentiator is positioning MSL scripts as high-quality academic discussions rather than sales pitches, with the Milton model reduced to academic rhetoric. Applicable scenarios include the design of daily MSL visit scripts, KOL insight and medical information communication, the construction of script systems for new product launches, stratified communication based on physician cognitive stages, and the generation of in-depth DA interpretation scripts. requires_api_key=false; no API Key required. The rating given on the official page is 100000, with 813 downloads and 0 installs. This article analyzes the Skill by framework dimensions and clarifies that installation commands and configuration field details remain unverified and are not fabricated.
Read Moredsh-ai-shopping-assistant: Organizing Shopping Evidence into Structured Decisions
dsh-ai-shopping-assistant is a DeepSeek Harness plugin (2.0.1-dsh.2) maintained by ShenXuAkaEkstasis. It registers the tool shopping_decision_analyze and performs local deterministic decision-making on the premise that the agent has already collected JSON evidence: candidate ranking, hard-constraint elimination, confirmed vs. conditional pricing, merchant-review weighting, and source-quality risk. The plugin does not connect to the network and does not place orders; it requires Node ^22.19.0 or >=24. The installation command is dsh plugin add github:ShenXuAkaEkstasis/dsh-ai-shopping-assistant#v2.0.1-dsh.2.
Read MoreTencent Health - Miying - Multi-Disease/Lesion AI Analysis Based on Super-Widefield Fundus Photography: Integrating Super-Widefield Fundus Images into AI Analysis
This is an enterprise-source Skill on SkillHub, designated as @tencent-adm/ultra-wide-fundus-images-ai-analysis (version 1.0.0), published by the Tencent Healthcare Miying team and categorized as 'professional'. It is designed for scenarios requiring the upload of ultra-wide-field fundus images for AI diagnosis and analysis, consolidating multiple model pipelines into a single trigger entry point. The Skill supports four capabilities: probable diagnosis, sign classification, sign detection, and sign segmentation, with trigger words including ultra-wide fundus, ultra-wide AI, wide-field fundus analysis, ultra-wide fundus, and ultra-wide diagnosis. It is tagged with **requires_api_key=true**. The article describes what it is, its core capabilities, and the prerequisite of preparing an API Key before activation. Due to a failed scrape of the official page, details such as installation commands, return structures, image limitations, ratings, and download counts could not be verified online; users should refer to the official source for accurate information.
Read MoreDividend Low Volatility Strategy: The Skill of the Three-Layer Framework for Trend, Valuation, and Execution
The Dividend Low Volatility Strategy Skill breaks down judgment into three layers—trend, valuation zone, and execution—while monitoring leading state-owned enterprises (SOEs). This article verifies its positioning, source, version, and classification: it is community-published, maintained by user_db7424f5, categorized as professional with version 1.0.0, and tagged with requires_api_key=false. The documentation only outlines the three-layer framework and monitoring targets; it does not elaborate on trend indicators, valuation scopes, or trading signal rules, nor does it provide installation commands, configuration fields, or reproducible usage examples. Consequently, the installation and typical usage sections are marked as not provided in the documentation; no interface names or parameters have been speculated. Discrepancies exist between the documented metrics (100,000 ratings, 902 downloads, 0 installs) and the task clues (99,629; 805). This article does not adopt these figures and advises verifying real-time data on the official page. This skill is suitable for individuals interested in reviewing dividend low volatility targets in a fixed sequence and focusing on leading SOEs. However, the framework does not guarantee returns, and any changes to the definition of parameters in any of the layers will affect the conclusions.
Read Moredsh-inline-view-tabs: Merge conversation view tabs into the DeepSeek Harness top bar in one line
dsh-inline-view-tabs is a DSH client web plugin maintained by Shellexy. It injects styles via client.js to integrate session view tabs, such as "Conversation" and "Trajectory", from a separate row into the header, positioning them right after the title and before the right-side tool icons. This reduces the header height from approximately 76px to 48px. Installation involves installing the npm bundle dsh-inline-view-tabs on the DSH plugin page and refreshing the browser. Selectors rely on public DOM anchors such as header[data-window-drag], with no Host configuration options. Edge cases such as narrow screens or header replacement should be noted, as upstream DOM changes may cause the layout to fall back to a two-row structure. Released under the MIT license and tested with dsh 0.2.1-rc.5.
Read Moredsh-plugin-updater: Check and One-Click Update Installed Plugins in DSH
dsh-plugin-updater is a DSH community plugin (admin-security) maintained by shehuilaoliumang. It compares bundles installed from npm via plugin_manager within a profile against the latest versions in the registry, and provides single/batch updates, version ignoring, and optional auto-update functionality in the sidebar, the "Update Check" section of the plugin page, and the bundle detail page. The update scope covers npm third-party plugins. Settings and ignore entries are stored in ~/.dsh/plugin-updater/settings.json. The installation command is: plugin_manager action=install_bundle target=<absolute path to plugin directory>. The Host side requires a DSH restart. Documentation includes instructions on SemVer status, rollback on failure, restart procedures, and pnpm lockfile repair.
Read MoreJiJing Cloud Temu Product Search: Turning Temu Sourcing into a Filterable Query
An introduction to the "JiJing Cloud Temu Product Search" Skill: Released by Qingdao JiJing Network Technology Co., Ltd. (Version 1.0.0, Category: business-ops). It supports filtering products by keyword, trusted category ID, site, sales volume, sales amount, price, supply price, count of identical items, rating, and listing time. Additionally, it facilitates the determination of best-seller rankings, new product rankings, low-competition products, candidate product comparisons, price band analysis, sales trends, competition intensity, and selection opportunities. The article outlines the skill's capability boundaries, activation methods, and applicable scenarios, and provides the official address.
Read Moredsh-web-search-anysearch: Connect web_search to AnySearch and enable one-click switching in settings.
dsh-web-search-anysearch (sheen945) is a DeepSeek Harness web-enabled tool plugin. It registers the AnySearch provider on the Host side, calls AnySearch via `https://api.anysearch.com/mcp`, and maintains the web_search tool contract. The `cordis.patch.yml` file selects AnySearch by default, while `client.js` provides an AnySearch/DeepSeek official switch card under Settings → Plugins. It supports `plugin_manager install_bundle`, three configuration locations for keys, and rate-limit retry. Requires DSH 0.2.1-rc.2, MIT 1.2.1.
Read More2024 List Pricing Standards and Measurement Standards (Chongqing Edition): Leave Provision Consultation to AI
This article introduces the "2024 Bill of Quantities Pricing and Measurement Standards (Chongqing Edition)" Skill released by community developer user_3df8d273. Covering GB/T 50500-2024, nine specialized engineering quantity calculation standards, and the Chongqing JJYJ50-003-2026 implementation rules, this Skill provides consultancy for Chongqing projects regarding clauses, coding, features, measurement rules, pricing procedures, and version applicability. Responses clearly distinguish between standard text, clause explanations, expert interpretations, and AI application suggestions, with sources cited. The service costs 0.40 CNY per confirmed question, which includes necessary supplementary retrieval, clarification, and error correction for that specific query; new independent questions are charged separately upon confirmation. Labels include pricing_type=alipay and requires_api_key=false. The article details the coverage, core functions, installation and activation methods, typical query formats, applicable scenarios, and precautions.
Read MoreSkill for Creating WeChat Official Account Articles: A Six-Stage Process from Topic Selection to Draft Completion
This Skill breaks down WeChat Official Accounts long-form article creation into six stages: topic selection, title crafting, outline structuring, drafting, saving, and graphic prompt generation. Each stage has clear objectives and respective documentation, and upon entry, the Skill first informs the user of the current stage. In the titleing stage, it must generate 5-10 title options and await user confirmation. The finalized draft consistently falls between 2500-3000 characters, covering four categories: practical tutorials, insight pieces, story cases, and list-based compilations. Once completed, the article is automatically saved as a Markdown file named in the format of "YYYYMMDD_keyword.md", followed by the automatic generation of creative prompts for cover and infographic creation. The image generation API will be triggered only after user confirmation. The writing style leans toward a conversational tone with complete paragraph expressions. Installation follows the command: openclaw skills install @fengbabao0929/wechat-article-writer. The source originates from ClawHub, and fittingly, users in China can resort to skillhub.cn as a mirror.
Read Moredsh-audit: Pre-installation Audit and Runtime Monitoring for DeepSeek Harness Plugins
dsh-audit is an auditing plugin for DeepSeek Harness maintained by Shark00I (v1.0.0, MIT). It performs plain-text pre-installation checks on directories, archives, and npm specifications, assigning S–D risk ratings. Post-installation, it logs plugin activations/deactivations, sensitive tool calls, and token usage estimates. Data is stored in `$DSH_HOME/dsh-audit/`. It provides a Settings → Audit panel, a `plugin_audit` tool, and an HTTP API. Requires DSH >= 0.2.1-rc.0 and Node >= 20. The scores are for risk assessment reference only and do not substitute for manual source code review. Third-party plugins run with the permissions of the current process.
Read MorePersonaMirrorVenue: As Agent Add A Layer MBTI Sixteen Personality Type Deployment
Personality Mirror Pavilion is an Agent Skill based on the universal SKILL.md format, located at @user_6d677e88/mbti-persona-simulator, published by user_6d677e88, tagged with source: community, current version 1.0.2, category: content-creation. It transforms the 16 MBTI personalities into callable role-playing entries, where summoning a specific personality switches to that personality's cognitive patterns, speaking style, and emotional responses. It supports scenario simulation, personality debates, reverse personality guessing, and personality analysis cards. It runs in pure text mode, with the tag requires_api_key=false. The article introduces its positioning, verified capabilities, activation methods, and typical use cases, and reminds users that personality traits are style constraints rather than factual capabilities, and that operational costs depend on the host platform.
Read MoreIMA Knowledge Account Automated Operation System: Full-Process Automation from Topic Selection to Scheduled Publishing
The IMA Knowledge Account Auto-Operations System is a community-published Skill located at @user_f12a44b7/ima-autopost-skill (version 1.0.1, category: ai-agent). It links topic pool design → AI article generation → routing and publishing to the corresponding knowledge accounts by category → daily scheduled execution via cron into a single pipeline, and provides state management, checkpoint-based resume, and sandbox recovery so that daily runs can be interrupted and later continued. It is suitable for operating multiple IMA Knowledge Accounts when you want to distribute content by category and rely on scheduled publishing. It is not applicable to manual single-post publishing (use ima-knowledge or ima-note) or WeChat Official Account articles (use wechat-article). Tag: requires_api_key=true. To enable it, an API Key and a cron environment are required. No official one-click installation command is provided in the documentation; it must be imported into an Agent environment that supports SKILL.md. It can be triggered by phrases such as “自动发知识号” and “IMA 自动运营”.
Read Moredsh-docwright: Generate Re-verifiable Structured Documents with Pack and a Validation Loop
dsh-docwright is a DeepSeek Harness plugin maintained by sharework-ai (v0.4.2, MIT license). It adopts JSON as the single source of truth: packs define schemas, rules, and generation prompts, supporting multi-stage chained scenarios, parameter extraction, and confirmation gates, while providing tools such as `doc_packs`, `doc_generate`, `doc_extract_params`, and `doc_validate`. Generated artifacts are accompanied by replayable verification receipts; in the event of failure, feedback is provided to the model for revision until validation passes or the maximum number of rounds is reached. The installation command is `dsh plugin add @sharework-ai/dsh-docwright`; the Quickstart guide demonstrates how to use `doc_validate` to perform zero-LLM deterministic validation on the cosmic-plan example.
Read MoreMEAL Evaluator in the Public Welfare Sector: Integrating Monitoring, Evaluation, Accountability, and Learning into Projects
The Public Sector MEAL Evaluator is an Agent Skill based on the PM4NGOs MEAL DPro knowledge system, integrating the four pillars of Monitoring, Evaluation, Accountability, and Learning. It provides support for full-cycle MEAL system construction, logic model design, data collection and analysis, adaptive management, and evaluation report writing for development and philanthropic projects. This article introduces its positioning, core competencies, activation methods, typical use cases, and applicable boundaries, and provides its official address.
Read MoreIntegrating the “Team · Board” into DSH: dsh-team
dsh-team (shaot92-del) is a DSH client plugin that, in a single package, provides 12 board_* and 4 sh_* tools, a sidebar "Job Board" panel, and /api/board/ui/* routes. Since v1.1, the Host backend handles take→dispatch→attach and final-state write-back, requiring a standalone HTTP task board (default 127.0.0.1:8965). Based on the GitHub README, this article covers installation commands, cordis.patch.yml configuration, typical agent/GUI usage, and board endpoint requirements. MIT licensed, current version 1.1.10.
Read Moredsh-session-filter: DSH Cross-Workspace Activity Sidebar
dsh-session-filter is a DSH client plugin released by Shane0315. It provides a bell icon entry in the workspace title row and displays cross-workspace activities in the sidebar, including: needs attention, activity from the last seven days, and optional pinned and scheduled task groups. Data is sourced from sessions.list, workspaces.list, and uiSession.sessionStatus, without reading conversation content. It supports search, pagination, local localStorage read states (not synced with native unread counts), and bulk menu archiving. Installation is recommended via the GitHub Release tgz or github:Shane0315/dsh-session-filter. It has been verified on DSH Desktop 0.2.1-rc.5 and is released under the MIT license.
Read MoreIn-Depth Review of Academic Papers: Delegating the Peer Review Process to Agent Skill
Introducing the SkillHub community skill "Deep Full-Text Paper Review": paragraph-by-paragraph, sentence-by-sentence review, P0–P3 graded issue list (location / evidence chain / root cause / recommendations + original vs. rewritten comparison), cross-chapter consistency checks, web-based fact-checking, and a Chinese A4 PDF report. Supports four review modes — deep / verification / mixed / concise — and multi-round tracking (prior-round fix log table, escalation after N consecutive unmodified rounds, green/yellow/red readiness status). Includes five built-in scripts for paragraph number extraction, data anomaly scanning, report typesetting, and PDF validation. After review, it can continue into a revision plan and produce substantial rewritten output. Published by @user_4f14f06c, v1.0.0, source: community, category: education, tag: requires_api_key=false, score: 100000, downloads: 906. Trigger terms cover: review my paper, deep review, verify last round's changes, give me a concise version for my advisor.
Read MoreAmazon Negative Review Alert · Review Sentiment Monitoring: Monitor spikes in negative reviews and rating drops for subscribed ASINs
This is a community Skill that brings Amazon review monitoring and alerting directly into the conversation. Coordinates: @user_507a5056/review-alerts, version 1.5.2, category business-ops. Once an ASIN is subscribed, the server continuously monitors its reviews and automatically generates alerts when negative reviews surge or star ratings decline. With a single command within the session, users can read unread alerts and locate problematic reviews, and further generate root cause analysis for negative reviews along with reply suggestions. Beyond alerting, the package also provides full ARI review analysis capabilities: subscribing to ASINs and collecting reviews, browsing and filtering reviews, viewing star ratings, keywords, trends, and radar charts, generating VOC and in-depth insight reports, running product operations workflows, monitoring product changes, accessing a negative review workbench, performing industry benchmarking and category ranking, and exporting review CSVs along with reports in Markdown/HTML. Using this Skill requires a valid ARI API Key (ari_live_*). The official page could not be scraped, so specific installation commands and configuration fields have not been verified; it is recommended to double-check them yourself before use.
Read MoreDaily Financial Briefing: Bilingual Morning Report Generation for Financial Managers Before Market Open
This article introduces the community-sourced "Finance Daily Morning Briefing" Skill (Coordinate: @user_a18131e7/finance-daily-morning-briefing, Version 1.6.0, Category: Professional). Designed for financial manager scenarios, it is activated via trigger words (e.g., "generate morning briefing," "write morning briefing," "today's briefing," "financial morning briefing," "morning briefing") to generate a bilingual (Chinese-English) pre-market financial report. The data is not hardcoded; instead, it relies on three sources: the `wind-mcp-skill` for data acquisition, the `skill-my-fund-manager` for manager profiles, and data scraped from the central bank and statistics bureau websites. The information is aggregated and organized into a final draft. The official tag `requires_api_key=true` indicates a dependency on authenticated external interfaces; users must ensure that API keys are configured for all dependencies before use. Official documentation does not provide a unified installation command or outline configuration differences for Cursor, Codex CLI, and Claude Code; the verified activation method is loading via coordinate reference. Typical usage involves entering a trigger word in the conversation, confirming external dependencies are in place, and verifying figures and sources before distributing the generated report. It is suitable for financial managers, investment advisors, and financial content teams requiring daily reports. Limitations include a strong dependency on external APIs and website scraping—where data is subject to official release schedules—and the necessity for manual verification of the institutional outlook sections.
Read MoreDSH-Screenshot: WeChat-style screenshot annotation beside the DSH input box
dsh-screenshot is a DSH client plugin maintained by sfunction-guo that adds a screenshot frame button to the conversation input box exclusively on Windows. It allows users to freeze the screen, drag-select a region, and annotate in-place (shapes, brush, text, numbering, mosaic). Once completed, the PNG image is sent directly to the attachment area or saved locally. Installation is available via the plugin marketplace, `npm install @sfunction/dsh-screenshot`, or via local bundle integration for development. It is ideal for illustrating UI issues and applying local masking within the Harness. Please verify the source code, MIT license, and DSH engine version requirements before installation.
Read MoreBaZi Fate-Calculating Astrology: Systematizing Chart Casting, Fortune Inference, and Textualization into a Skill
This article introduces the community-sourced Skill "Bazi Fortune Telling & Destiny" (Coordinates: @user_b75ca509/bazi-fortune-telling-destiny, Version 1.8.8, Category: life-service, Tag: requires_api_key=false). Based on the user's ID card or Gregorian birth date, birth time, and gender, the Skill casts the "Four Pillars" (Bazi). It utilizes systems such as the Five Elements (Wu Xing) generation and control, Yin-Yang theory, the Ten Gods, relationships between the Twelve Earthly Branches (Six Harmony, Six Clash, Triple Combination, Punishment, and Harm), and Major Luck and Annual Cycles (Da Yun Liu Nian) to calculate the overall destiny. It highlights potential timings for significant wealth and major calamities, offering methods to facilitate or mitigate these outcomes, ultimately generating a standard destiny document. The text outlines input requirements, trigger words, the fixed pipeline, and typical conversation patterns, while noting that accuracy depends on input validation and that Mingli deductions should not serve as the sole basis for serious decision-making. For installation, please refer to the official page; no speculative commands are provided in this article.
Read MorePrivate WeChat Moments Content Operations Officer | Jianshi AI: Turning Moments Content into an Actionable Content System
This article introduces the community Skill on SkillHub, "Private Domain Moments Content Operator | Jianshi AI." It generates a directly executable content system for Moments, WeChat private domain content, and customer education. The process involves first clarifying the target audience and the specific concerns to be addressed, then scheduling a calendar of 30 content items, designing opening hooks and calls to action, generating copy categorized as problem-oriented, educational, evidence-based, boundary-setting, or invitation-based, establishing a publishing rhythm, and identifying effective interactions or reviewing content with low consultation rates. The publisher is user_fccf428b, with a category of content-creation and version 1.0.5. The tag requires_api_key=false indicates that no additional API Key is needed. The provided materials do not include installation commands or configuration fields, only the coordinate @user_fccf428b/private-domain-content-calendar; therefore, this article describes the activation method based on standard SKILL.md practices and advises users to verify independently. Additionally, it should be noted that there is a discrepancy between the leads (rating 99752, downloads 806) and the retrieved materials (rating 100000, downloads 866); this article prioritizes the retrieved data.
Read Moredsh-model-pin: Pinning DSH New Session Default Model
dsh-model-pin is a DSH Cordis plugin maintained by SeyanaA. It replaces the built-in agentDefaultModel so that new sessions always use the configured provider, model, and optional reasoningEffort. The in-session model selector and /model remain available but do not write back to the default. It supports configuration via the settings page and cordis.patch.yml, volatile hot-reloading, and periodically checks (based on modelCheckInterval) for new provider models. It is primarily used together with Codex-to-DSH-Bridge. Compatible with DSH Desktop 0.2.x, registered to the profile via the local link directory.
Read More