FDE Intelligent Workbench Expert Team: Placing Enterprise AI Project Delivery into a Propelling Battlefield
The FDE Smart Workbench Expert Group is an enterprise AI project delivery agent platform published by SkillHub user user_30f2bcde (version 2.4.1, category: ai-agent). It organizes customer diagnosis, solution workshops, deployment guidance, value calculation, knowledge assets, and decision scheduling into six warrooms spanning 24 subpages. The interface provides global search, a command palette (Ctrl/⌘+K), multi-project parallelism, data backup/restore, and a presentation mode, and supports on-demand one-click calls to sub-agents, a four-state confirmation gate for results, cross-warroom data lineage linkage, and export to Markdown or print to PDF. In direct chat, interactions progress step by step via stage cards and choice cards rather than throwing a questionnaire all at once. This article describes its positioning, verified features, items to confirm before enabling it, and typical usage scenarios.
Read MoreIndex Valuation: Nasdaq 100 / S&P 500 / Dividend Low Volatility Valuation Daily Report Generation Skill
"Index Valuation" is a SKILL.md-format skill released by the SkillHub community. Version 1.0.1, categorized under content-creation, and requires no API key. It scrapes indicators from three data sources—multpl.com (S&P 500 valuation history from 1871 to present, 1,867 monthly records), Danjuan Fund (real-time PE/PB/percentile/dividend yield/ROE for 63 indices), and Sina Finance (closing levels for the S&P 500, NASDAQ-100, and NASDAQ Composite)—and calculates PE/CAPE/PB/dividend yield/PS percentiles over the past 10 years, the past 20 years, and the full history. It outputs an HTML daily report card (with the NASDAQ-100 pinned to the top and including a brief “Today at a Glance” guide) and a text brief that can be directly copied to Xueqiu, WeChat Official Accounts, and Xiaohongshu. The skill only presents historical percentile data, provides no directional judgments or buy/sell advice, and includes a disclaimer. It is suitable for daily index valuation tracking, generating finance content for self-media, and referencing the historical position of market valuations.
Read Moredsh-agent-team-model-selection: Specifying LLM Routing for Agent Team Members
The DSH community plugin `dsh-agent-team-model-selection` replaces the official Agent Teams experimental tool list, adding support for `provider`, `model`, and `reasoning_effort` in `spawn_teammate`, and providing `llm` fields for `list_team_models` and `list_agents`. The settings page supports two modes: `inherit` (automatically follows the Lead, with an optional whitelist) and `fixed` (unified routing), which can also be configured via `cordis.patch.yml`. Install using `plugin_manager install_bundle` with the target `devloom1024/dsh-agent-team-model-selection`. Released under the MIT license, current version 0.6.0.
Read MoreAmazon Ads: Account Health Check Tool for Amazon Sellers
"Amazon Ads" (amazon-ads-diagnosis) is an ad audit Skill published on the SkillHub community, maintained by user_57b05cf6 (Version 1.0.0, Category: business-ops). Designed for Amazon sellers, it rapidly diagnoses ad account health across four key dimensions: AI vs. manual ad performance comparison, wasteful spend analysis (non-converting keywords/competitor ASINs), high-quality traffic identification (low-ACOS keywords/competitor ASINs), and ACOS anomaly troubleshooting. It generates a structured diagnostic report to help sellers pinpoint where their ad budget is going, identify areas of waste, and determine which traffic sources warrant further investment. This Skill is limited to diagnosis and analysis; it is free to use, requires no API key, and does not perform price or budget adjustments. For automated optimization, please use the "Eight Elephant AI" (八头象) ad optimization skill (btx-ai-ads).
Read Moredsh-multi-shell: One-time Shell Tool Registration, Choose Git Bash / PowerShell / WSL per Execution
dsh-multi-shell is a DSH workflow plugin maintained by deepseekv41flash that registers a single shell tool. This tool allows the model to select the backend on each invocation via a `shell` parameter, choosing from Git Bash (default), PowerShell, or WSL. Its key difference from `dsh-bash-terminal` is that the backend is specified per-call by the model, and its `peerDependencies` use a broad range to reduce version-gating for host upgrades. The article introduces the tool parameters, the argv for the three backends, sandbox and approval semantics, and profile configuration options. It also covers the npm installation command `dsh plugin --profile desktop add dsh-multi-shell` and considerations for bundle registration. Finally, it reminds users that the plugin is mutually exclusive with shell plugins of the same name and that commands are executed with the permissions of the DSH process.
Read MoreTikTok Operations: Integrating the Douyin Operations SOP into an Agent's Skill
Agent-oriented Douyin operations guide and SOP Skill, published by clawkk on ClawHub. Version 0.1.0 is released under the MIT-0 license and has passed security audits. It embeds four core operational guidelines: a content cadence of 3-5 videos per week, recommendation mechanisms driven by completion and engagement rates, e-commerce monetization chains (product cards, commissions, showcases), and compliance red lines (prohibiting exaggerated claims and mandating ad disclosures). It can be installed via `openclaw skills install @clawkk/douyin-operations` without an API key. As a guide-based Skill, it provides no data APIs and does not automate backend operations. Updates to platform rules must be manually synchronized, and its integration compatibility with other editors has not been verified.
Read MorePrelude: DSH Prompt Optimization and Rewriting Assistant
Prelude (dsh-prompt-coach) is a DSH community plugin maintained by dbcaskml183-dotcom, offering 90 method cards, prompt analysis, sentence insertion, and full-document rewriting with reference to the original manuscript. Browsing the method library requires no model; analysis and rewriting use DSH's configured models. Supports DSH 0.2.1-rc.2 to rc.5 and Node 24+. Install via `dsh plugin --profile web add dsh-prompt-coach@0.4.24 --ignore-scripts`, then open Prelude next to the input box to use. MIT licensed, independent community project.
Read MoreFirst-Class Registered Architect AI Assistant: 7 Roles Covering Exam Preparation and Professional Practice
The Level 1 Registered Architect AI Assistant (@hongnan/zhumengshi-architect-team) is released by Guizhou Hongnan Network Technology Co., Ltd. (Version 1.1.0, Category: Professional, Source: Enterprise). With a page rating of 100000 and 1259 downloads (0 installations), its tag `requires_api_key=false` indicates that no API key is required. This Skill divides the Level 1 Registered Architect context into 7 roles: six exam preparation coaches (Knowledge Organizer, Question Generator, Construction Instructor, Structural Calculator, Progress Manager, and Drawing Coach) and one professional application advisor (Code Compliance Consultant). Its capabilities cover two main areas: pre-exam coaching (systematic explanations, key/difficult points, high-frequency exam topics, practice drills, and drawing strategies) and professional application (applicability of code provisions, scheme design compliance, construction drawing review essentials, and registration practice). These roles can be invoked individually or collaborated to form a workflow. As the retrieved information does not provide installation commands or configuration fields, this article describes the facts as they are. It is recommended to load the Skill in a supported client following the general SKILL.md method; please refer to the official page for specific details.
Read Moredsh-plugin-effort-panel: Single-Layer Panel for Codex-Style Models and Reasoning Effort in Composer
dsh-plugin-effort-panel consolidates model selection and reasoning effort in the DSH 0.2.1 series Composer into a Codex-style single panel: the upper half selects the model, while the lower half provides a slider to adjust the effort level, with state kept in sync with /model and modelDirectories. It supports drag-to-snap, token statistics, and balance display after login, and installation requires a full restart of DSH. Based on the official README and package.json, this article provides desktop/web installation commands, compatibility scope, and known limitations.
Read MoreMinimax H3 Video Generation and Editing: AIGC Video Skill Powered by Text, Image, and Audio
Minimax H3 Video Generation and Editing is an Agent Skill based on SKILL.md, located at @infimind/minimax-h3-video-generation-and-editing. Version 1.0.0, released by JIRI Technology - AI Expert and categorized under design-media, it targets video editors, post-production teams, advertisers, and creators requiring asset reuse. It supports generating content from scratch using text alone, or controlling results with images, video, or audio inputs, covering text-to-video, image-to-video, first-last frame animation, reference-to-video, video-to-video, and audio-to-video. When used via AI Hive, it automatically uploads assets before generation, and upon submission, automatically saves tasks, tracks progress, and downloads the final video. It is suitable for AI advertising, TVCs, e-commerce videos, product showcases, live commerce, content seeding, short dramas, motion comics, and social media content, serving as a functional alternative to similar capabilities in tools like Kling, Jimeng, Hailuo, Vidu, PixVerse, Runway, Sora, Veo, and HeyGen. Tagged with requires_api_key=true, it requires API Key configuration. As installation commands and configuration fields were not provided in the scraped data, they are not fabricated here; please refer to the official page for details.
Read MoreOfficial Account Subscription: Aggregate your subscribed articles daily at 6 AM
"WeChat Official Account Works Subscription" is a Skill for subscribing to WeChat Official Account articles, published by SkillHub community user user_5f9c21aa. Currently at version 1.0.2 and categorized under data-analysis, it requires an API Key. It allows users to subscribe to competitors, similar accounts, and specific Official Accounts of interest, automatically generating a daily digest of articles from these accounts at 6:00 AM. This automates the repetitive task of manually checking Official Accounts daily. This article covers its positioning, verified capabilities, official address, prerequisites, typical usage, and applicable scenarios. The page materials do not provide a unified installation command; loading methods vary by client, so please refer to the official page for instructions. As this Skill is community-sourced, interface stability depends on the publisher's maintenance, and the scraping of Official Account articles is subject to WeChat platform policies.
Read Moredsh-onebot-channel: Integrating QQ Private and Group Chats into DeepSeek Harness via OneBot 11
`dsh-onebot-channel` is a DSH client plugin maintained by danshuyue, built on OneBot 11. It features WebSocket inbound connections, route-to-session mapping, and turn triggering via `agent.followup`, while providing ten outbound tools such as `qq_send`. The installation command is `dsh plugin --profile desktop add dsh-onebot-channel`, with configuration stored in `~/.dsh/qq-bridge.config.json`. It supports private message merging, group @ context, quotes and @ mentions, relationship memory, and optional proactive message initiation. Users must provide their own OneBot protocol endpoint; the plugin runs with the permissions of the `dsh` process and is licensed under MIT.
Read Moredsh-prompt-snippets: Reusable Prompt Snippet Library and One-click Insertion
The DSH client plugin dsh-prompt-snippets, maintained by dadazhangn, stores reusable prompts locally in snippets.json, supporting categories, tags, favorites, version history, and crash-safe atomic writes. {{placeholders}} can trigger form dialogs or insert default values directly, with built-in date/time macros; an input dock panel and Ctrl+Shift+K global search enable one-click insertion. No data is uploaded to the network by default, except when explicitly opted in for AI optimization. Installation: dsh plugin --profile web add dsh-prompt-snippets; requires DSH >= 0.2.1-rc.3.
Read MoreAI Compliance Expert Team for Financial and Tax Intermediaries: Breaking Down Compliance Consultations into Thematic Experts
When delivering compliance services, tax and accounting firms often face complex client requirements involving contract tax clause review, legal-tax intersection analysis, and legal review. General conversational models often suffer from singular perspectives and opaque sourcing. The "AI Compliance Expert Team for Tax and Accounting Agencies" introduced in this article is a compliance consulting Skill designed for tax and accounting intermediary agencies. It is located at @user_11064e10/tax-expert-firm-ai-compliance, published by user_11064e10, sourced from the community, classified as professional, currently at version 3.18.64, and tagged with requires_api_key=false. It utilizes the firm's supply-side director as a coordinator to orchestrate specialized experts in intelligent compliance consultation, contract tax review, joint legal-tax review, and legal review, with the goal of enhancing the compliance service capabilities of tax and accounting firms. The text outlines capability boundaries, the verifiable configuration of requiring no API key, the SKILL.md format loading method, and the official address. Simultaneously, it objectively notes that the retrieved materials do not provide official installation commands, directory paths, or invocation examples, and the official page failed to load during this retrieval. It suggests independent verification prior to deployment and manual review of any compliance opinions to be issued.
Read Moredsh-memory: Distilling Long Conversations into Reusable Skills and Negative Knowledge
dsh-memory is a DSH ecosystem memory plugin maintained by corvus-dsh: it reads the handover ledgers and sessions of dsh-graft-slice, writes reusable practices from sessions into SKILL.md and troubleshooting paths into negatives.jsonl, and retrieves and injects them according to workspace and task alignment rules before new tasks. It supports LLM extraction and semantic filtering, bi-temporal and supersession auditing, negative knowledge validity and skill archiving, and provides a Web memory console. Install and use it via the dsh plugin add command and cordis.patch.yml configuration in the README; the plugin runs with host process permissions and requires Node 20+ and DSH >=0.2.1-rc.5.
Read MoreWeCom Moments Content Data Review and Improvement Sheet: Turning Performance Gaps into Next-Round Improvements
This article introduces the community-sourced Skill "WeCom Moments Content Data Retrospective & Improvement Sheet | Jianshi AI": coordinates @user_fccf428b/wecom-moments-content-retrospective, version 1.0.0, category business-ops, requires_api_key=false. It addresses the gap in community operations where "data is seen but not translated into the next round of actions": it identifies differences in the performance of Moments content, turns them into specific improvements for the next round, delivers a content improvement sheet, and rechecks using the Moments effective response rate. The article provides a breakdown of core functions, available activation information, and a four-step usage flow (prepare data, trigger the retrospective, obtain the improvement sheet, and recheck via the effective response rate), and describes the target audience and limitations: it does not collect data from the WeCom admin console, does not generate new copy, and the definition of the effective response rate must be confirmed by the user. Because the official page was unavailable at verification time, the installation command is not included in the main text to avoid providing unverified information.
Read Moredsh-graft-slice: Physical Slicing of Long Sessions and Zero-Inheritance Cold Recovery
dsh-graft-slice is a DSH memory-class plugin maintained by corvus-dsh. It slices out a high-fidelity tape at the final turn/end of a source session, creating a target session with zero inheritance that can be cold-recovered—no forking, no lossy compression—with crash safety guaranteed by transaction logs and ledger epochs. It supports handoff task boards, semantic slimming, physical ignition, and seq-based read-back of graft_read_result/graft_search_history. Installation: dsh plugin --profile <name> add dsh-graft-slice; requires DSH >= 0.2.1-rc.5.
Read MoreGEO First Principles: Assessing Necessity Before Outlining a 10-Step Action Roadmap
GEO First Principles (@user_4a9e056e/geo-first-principles) is a GEO decision analysis Skill maintained by community developers. Categorized under business-ops (v1.1.0), it does not promise rankings. Instead, it analyzes customer purchasing behavior, industry competition, and business objectives to determine a company's suitability for GEO. It generates an HTML report featuring a left-side navigation, chapter illustrations, a six-part analysis, five self-assessment questions, and a ten-step action roadmap. No API Key is required to run it. It is designed for enterprises to evaluate GEO investments, teams to create internal assessment materials, and service providers to conduct preliminary diagnostics. Report conclusions depend on input quality; the tool does not perform automated testing nor guarantee recommendations or sales conversions.
Read Moredsh-plugin-codex-link: Driving Codex Sessions in DSH
dsh-plugin-codex-link is a DSH community plugin maintained by CN-Mg. Leveraging Codex's built-in app-server and codex exec, it enables switching Codex sessions, sending commands, reading processes via codex_watch, and dispatching sub-agents via codex_task within DSH, while allowing mode switching between direct control, sub-agent, and hybrid in the settings page. No plugin installation is required on the Codex side; it can be installed via `dsh plugin --profile web add dsh-plugin-codex-link` or by entering the package name in the desktop app. It includes documentation on binding persistence, quota queries, queue fallback, and cordis configuration options, suitable for developers integrating DSH with Codex.
Read MoreAnyAudio: Empowering Agents to Turn Sound Ideas into Remixable Audio
AnyAudio (dsh-any-audio) is a DSH client plugin maintained by cloudoyx, providing AI agents with sampling instrument rendering, millisecond-level timeline orchestration, sound effects, Edge TTS and MiMo-V2.5-TTS narration, multi-track mixing, and video audio track synthesis. To install, use the command dsh plugin --profile desktop add dsh-any-audio. Audio sources need to be configured in the AnyAudio page under Settings, with SoundFonts placed in ~/.dsh/anyaudio/fonts. MIT licensed, early release, requires Node 22.19+ and ffmpeg (for speech/video).
Read MorePinduoduo E-commerce Video Generation and Editing: AI Video Skill for E-commerce Assets
This is an Enterprise Skill published by G-Byte AI Expert, categorized under design-media. Located at @infimind/pinduoduo-ecommerce-video-generation-editing (Version 1.0.0), it is designed for e-commerce operators, product teams, and content creators. It supports text-to-video generation and allows the incorporation of images, videos, or audio to control the output. Features include keyframe animation, reference-based generation, video-to-video translation, audio-to-video synthesis, and AI video editing and extension. When used with AI Hive, it automatically uploads required assets before generation and automatically saves tasks, tracks progress, and downloads the final video upon submission. It is ideal for AI advertising, TVCs, livestream commerce, product seeding, short dramas, comic dramas, and social media content. Tag: requires_api_key=true; an API Key must be prepared before use.
Read MoreSQL Optimizer: Auditing SQL for performance and correctness to produce optimized queries
sql-optimizer is a community-contributed "dev-programming" Skill on the SkillHub platform, located at @user_aa378cba/sql-optimizer. Version 1.0.0 requires no API Key. It is designed to conduct performance and correctness reviews on one or more SQL queries, producing actionable optimized SQL and a analysis report. On the performance side, it covers slow query rewriting, redundant join removal, and index recommendations; on the correctness side, it addresses common pitfalls such as row multiplication, NULL handling, and type mismatches. It automatically identifies four dialects—SQL Server, MySQL, Oracle, and PostgreSQL—and provides index and syntax recommendations tailored to the specific dialect. To use it, provide a .sql file or paste SQL directly, and trigger the process by requesting "optimization," "tuning," "review," "speed up," "add index," or "rewrite." Official documentation does not provide specific installation commands or reproducible examples; integration methods should refer to the official page. It is suitable for scenarios such as batch reviews of existing SQL, slow query rewriting, and troubleshooting abnormal result sets; however, the generated recommendations still require verification on actual databases.
Read Moredsh-model-provider-picker: Vendor-first Two-level Model Selection and Visibility Control
dsh-model-provider-picker is a DSH profile plugin maintained by cjc-github that changes the composer model selection flow to select vendor first, then model. Models with reasoning levels proceed to a second screen for level selection. The settings panel provides vendor/model hide toggles, with rules written to the Host's model-provider-picker namespace. Installation requires adding to dsh.profile.bundles via pnpm link or tgz in the profile directory, and a Harness restart to take effect. It only affects the Web/desktop GUI and depends on the official model-selection and modelDirectories.
Read MoreCinematic Shooting and Camera Movements: An Introduction to the Skill for Storyboarding Video Content
This article introduces the community Skill 'Cinematic Shooting & Camera Movements' (@user_e655b91e/video-cinematography, v1.0.2, categorized under design-media). The skill is triggered when users mention camera movements, lensing, shot types, push/pull/pan/tilt shots, cinematic framing, or when they need to design storyboard scripts, providing specific suggestions at the camera language level. The article verifies the trigger conditions, capability scope, source attribution, and page address. It truthfully states that the obtained materials do not provide installation commands, configuration fields, or output examples, and does not infer unverified content. It recommends activating the skill following the official page guidance.
Read Moredeepinfra-fusion-proxy: A Local Reverse Proxy Injecting service_tier and Standard↔Flex Fallback for DeepInfra
deepinfra-fusion-proxy is a dependency-free local reverse proxy maintained by citydirector, designed for DeepInfra's OpenAI-compatible API. It injects `service_tier` and `fail_fast` into chat completions, performs per-session routing based on `prompt_cache_key`, and automatically retries with Standard when encountering 429 errors on Flex, supporting cooldown locks and Composer switching. DSH ≥0.2.1-rc.2 and Node ≥24.9 users can enable it via Plugins by adding the package name `deepinfra-fusion-proxy` or using the CLI command `dsh plugin --profile web add`. It listens on 127.0.0.1:8790 by default; providers must set `baseURL` to the local address and use `cacheRetention: long`. Licensed under MIT, the source code is also available on SkillHub and GitHub.
Read More