Organization of Reimbursement Invoices and Generation of Reimbursement Forms: Consolidating Scattered Invoices into Standardized Reimbursement Drafts
At month-end reimbursement, invoices are often scattered across photo albums, emails, and paper slips, making manual classification, summation, and de-duplication both slow and error-prone. The "Reimbursement Invoice Organizer and Form Generator," maintained by SkillHub community user user_af5149c6 (Version 1.0.3), falls under the office-efficiency category with the tag requires_api_key=false. It compiles image or PDF invoices (processed via AI recognition) or direct lists into a categorized reimbursement draft, a ticket attachment list, and reminders for items requiring verification. The tool supports automatic classification into five categories (transportation, dining, accommodation, office, and communication), detects duplicate invoices, calculates totals, and generates capitalized RMB figures. It is suitable for personal travel reimbursements, daily expense claims, and monthly expense consolidation. This article outlines its capabilities and limitations across sections including "What is this," "Core Features," "Installation & Activation," "Typical Usage," and "Use Cases & Notes," while clarifying that the official webpage could not be scraped and the installation command remains unverified.
Read MoreAITP Research Workbench: Constraining Subject Catalogs and Write Locations for DeepSeek Harness
AITP Research Desk is a DeepSeek Harness client plugin (v0.1.0, MIT) maintained by bhjia-phys. It defines folders containing `research.md` as research topics, constrains agent write locations via tools such as `research_path`, write guards, and `research_check`, and provides a file manager-style topic tree, planning pages, an editor, and a knowledge base browser. It is recommended to install this alongside the AITP-Research-Protocol Skills bundle to share `placement-rules.json`. Installation requires DSH 0.2.1-rc.2+ and Node 22+. To enable it, run `dsh plugin add` with the local path after cloning, and configure `roots` and `placementRules` in `cordis.patch.yml`.
Read MoreReading Notes Mind Map Generator: Condense a Book into a Structured Mind Map Outline
This article introduces the community-sourced Agent Skill "Book Note Mind Map Generator" (@user_32cf4d2a/book-note-mindmap, knowledge-management, v1.0.0). By inputting a book title and core content, it outputs a structured full-book mind map outline covering core viewpoints, key cases, golden quotes, and action suggestions. The article verifies its functionality, source attribution, and page snapshot data (rating 100000, downloads 1076, installs 0, requires_api_key=false), and provides an input template and applicable scenarios. It honestly notes that the official installation command and SKILL.md fields were missing from the retrieved data, without fabricating any details.
Read MoreThe AI Reading Crew: Hosting a Multi-Expert Book Club in Any Chatbot
Reading Squad is a community Skill designed for AI Agents (Version 1.0.0, Source: community, Category: ai-agent), published by user_3a25e955. Based on the universal SKILL.md format, it has no platform dependencies and requires no API key (requires_api_key=false). It addresses the challenges of digesting a book from multiple perspectives after reading and the difficulty of organizing real-world book clubs. It creates a "Squad" composed of multiple expert Agents with distinct personas. Under the centralized orchestration of a Leader, the squad engages in multi-round discussions centered on a specific book or topic to synthesize a consensus. Core capabilities include member definition, leader orchestration, multi-round discussion, and consensus synthesis. Since official documentation does not provide installation commands or configuration fields, no inferences have been made in this summary; readers are advised to verify details on the official page. It is applicable to scenarios such as multi-angle reading digestion, virtual book clubs, and multi-perspective viewpoint rehearsals. Limitations include the fact that discussion quality depends on the specificity of inputs, the output represents AI simulation rather than real expert opinions, and with zero installations, available reference information is primarily derived from the official description.
Read MoreBilibili Video Summarization and Key Point Extraction: Converting Public Videos into Structured Summaries
Bilibili Video Summarization and Key Point Extraction is a community Agent Skill based on the SKILL.md format, maintained by user_113322ce, Version 1.0.2, Category: knowledge-management. It is designed for developers and content creators who need to quickly digest public videos on Bilibili. Given a video link and a summarization request, it outputs a structured result consisting of five parts: a one-sentence summary, a content overview, key viewpoints, key details, and action items. No API key configuration is required. It does not output verbatim subtitles, does not perform account data analysis, and does not generate videos. If a link is inaccessible or subtitles are missing, it reports the issue as-is without fabricating content. This article describes its positioning, core features, triggering mechanism, use cases, and limitations.
Read Moredsh-split-reply: Send QQ Channel Replies Individually by Sentence
dsh-split-reply is a DeepSeek Harness plugin maintained by betty598. It wraps the sendMarkdown function of dsh-qqbot: it splits messages into sentences based on punctuation and line breaks, then sends them one by one with random intervals of 0.7–1.6 seconds, simulating human typing. You need to disable streaming in im-qqbot and configure items such as maxSegments. It only works in QQ channels and is subject to the passive reply limit. Licensed under MIT, the current version is 0.1.1, featuring reload self-healing.
Read MoreWeekly Report Generation Skill: Summarize Git commits by author to generate Chinese weekly reports.
The Weekly Report Summary Skill (@user_fff6db65/git-weekly-report-skill, v1.0.0, categorized as office-efficiency) scans all Git repositories within a specified directory, analyzes commits by time range, aggregates them by contributor, and generates Chinese weekly reports. Capabilities include counting commits from the past week, estimating task duration, allocating work hours in 0.5-day increments, and normalizing total hours to a default 5-day work week. It can be triggered via natural language in both Chinese and English, with the tag requires_api_key=false indicating no API key is needed. This document describes the core features, official repository URL, activation methods, and applicable scenarios, while noting usage limitations such as unverified installation commands, commit attribution issues, and time range constraints.
Read MoreDSH-Sync-Updater: Unified Inventory, Check, and Update for Skills and Plugins in DSH
By default, DSH does not proactively pull skill and plugin upstreams. dsh-sync-updater, maintained by Bertli486, provides a sidebar "Sync" panel, three model tools (sync_status/sync_check/sync_apply), and configurable scheduled checks; it supports inventorying local skills and profile dependencies, comparing registry entries with SKILL.md hashes, applying updates, and installing new skills via skills.sh from the discovery page (including security audit). Installation command: dsh plugin --profile web add github:Bertli486/dsh-sync-updater. MIT licensed, categorized under admin-security.
Read MoreTraining Materials Assessment Skill: Generate Assessment Questions from Training Documents along with Key Answer Points
Training Material Quiz Generator is a community Skill based on SKILL.md, located at @user_bd3231bd/training-quiz-from-doc, version 0.1.0, categorized under education, and tagged with requires_api_key=false. It currently has a rating of 100000, 1083 downloads, and 0 installations. It has a very narrow focus: it takes training materials as input and outputs post-training assessment questions along with key points for each answer; it does not generate training outlines or FAQs. The article introduces what it is, its core functions, installation and activation (this time the official page crawl failed, so it can only be confirmed that no API Key is required; if no installation command can be verified, it will not be fabricated), a typical use case by invoking it via its identifier, applicable scenarios and caveats (question quality depends on the quality of the materials, answer key points need graders to supplement scoring criteria, the version is new and has zero installations), and concludes with the official URL.
Read Moredsh-shortcut-extensions: Adding a layer of keyboard-driven quick switching to the DSH interface
dsh-shortcut-extensions is a DSH client plugin maintained by BenHanks303. It provides a model and session palette, session cycling, jumping to answered unread sessions, Composer focus, and configurable permission cycling on both Web and Desktop platforms. It also includes a sidebar control panel and a keyboard shortcut reference. Released under the MIT license, version 1.1.0, written in pure JavaScript with no build step required. Installation involves cloning the repository, then installing from the local directory via the DSH plugin page, followed by reloading the window. This plugin is ideal for keyboard-driven workflows and multi-session users. Please note that browser keyboard shortcuts are reserved, and the plugin runs with the permissions of the current DSH process.
Read MoreSimulated Bid Evaluation · Reverse Engineering the Evaluation Sheet: Extracting Three Key Takeaways
# Introduction Bidders often face a common pain point: receiving the "Bid Evaluation Table" within tender documents reveals a mass of scoring criteria, point values, and rejection clauses, yet it remains unclear how many points their response materials will actually secure, where deductions will occur, or where they might result in direct disqualification. Manual clause-by-clause comparison is time-consuming and laborious, and prone to missing hidden point losses and formal defects. The "Bid Evaluation Simulation · Reverse Engineering of Evaluation Table" Skill introduced below addresses this issue. Standing on the bidder's side, it reverse-engineers an evaluation table into executable scoring, navigation, and compliance review results. # What is this A three-in-one analysis Skill for the bidder side, located at @user_7ae399a8/bid-mock-evaluation, published by user_7ae399a8, sourced from commun.
Read Moredsh-thinking-intensity: Replacing Input Field Model Tier Selection with Thinking Intensity Capsules
dsh-thinking-intensity (dsh-thinking-pill) is a DSH Web Profile plugin maintained by BenedictTom that takes over the input box model seat, displays and switches reasoning.efforts in the model directory using light point gear names and particle flow, and writes back to the session through modelDirectories.select. This article introduces its gear bindings, visual rules, installation command dsh plugin add BenedictTom/dsh-thinking-intensity, as well as notes on local link development and usage.
Read MoreFinancial Report Generation: Convergence of Four Reports into a Single Skill
This article introduces the SkillHub skill "Financial Report Generation" (user_3c6cb52e/financial-report-generation-skillhub, v1.0.0, community source, professional category, published and maintained by user_3c6cb52e). Targeting development scenarios that require consistent production of financial reports, its official capability statement describes generating balance sheets, cash flow statements, income statements, and executive financial summaries, thereby reducing reliance on ad-hoc scripts for template assembly and mitigating inconsistencies in output standards. According to collected data, the skill has a rating of 100000, 1078 downloads, and 0 installations, and is tagged with requires_api_key=false. The article explains what it is, core features, installation and activation (no verifiable commands are provided in the source; instead, point to the official page, remind readers to confirm download and activation methods first, and note that differences among Cursor / Codex CLI / Claude Code cannot be verified from the available materials and are therefore not covered), typical usage (input account balances or trial balances; generate reports in the order of balance sheet, cash flow, income statement, and executive summary while maintaining consistent standards), applicable scenarios and precautions (the source does not provide compliance standards or audit rules; formal disclosures require review; with 0 installations, validate with test data first; when fixed tax forms or accounting standards are required, clarify the applicable standards in advance). The official address is https://www.skillhub.cn/skills/user_3c6cb52e/financial-report-generation-skillhub
Read MoreEnergy Consumption Analysis Report: Organizing Building Energy Data into a Deliverable Report
The "Energy Consumption Analysis Report" is a Skill designed for generating reports based on building energy consumption data. Identified as @user_ffc4b166/energydata-analysis-v1 (Version 1.0.0), it originates from the community, is published by user_ffc4b166, falls under the data-analysis category, and does not require an API Key. It is designed to be automatically triggered by keywords such as building energy consumption, energy analysis, carbon emission calculation, energy consumption reports, energy-saving analysis, building energy management, energy audits, carbon emission indicators, energy consumption per unit area, building energy efficiency standards, and energy consumption data tables. It should also be activated even if a user simply states, "Help me analyze the energy consumption data for this building." The input supports Excel and CSV files or direct pastes of tabular text. The output comprises four modules: energy ratio analysis, monthly energy usage trends, carbon emission calculations, and benchmarking against building energy efficiency standards. This article outlines the Skill's positioning, trigger mechanisms, core capabilities, reproducible typical usage, and applicable scope. It also highlights that the captured documentation lacks installation commands and the specific caliber for emission factors. Consequently, it is recommended to validate the tool with sample data prior to utilizing it for official deliverables.
Read Moredsh-fakeip-compat: Allowing DSH web_fetch to Accept FakeIP Range DNS Answers
This article introduces the community plugin dsh-fakeip-compat (by bemarkt). When a local FakeIP DNS returns placeholder addresses such as 198.18.0.0/15, the built-in web_fetch of DeepSeek Harness rejects requests due to "non-public IP" restrictions. This plugin inserts FakeIP range allowlisting rules while preserving existing IP pinning, redirects, and limits. It also offers an optional DNS-over-HTTPS (DoH) check to verify the real destination. The article covers environment requirements (DSH 0.2.1-rc.5, Node 20+), installation via plugin_manager, bundle patching, General settings configuration, and the rationale for keeping DoH disabled by default.
Read MoreFeishu CLI Setup: Turning Natural Language into Lark CLI Commands
Lark CLI Installation & Configuration is a community skill for Agent Skills (ID: @user_d8831e99/lark-setup, Version 1.0.0, Category: dev-programming). It handles two main tasks: guiding users through the download and installation of lark-cli, and translating natural language requests into corresponding lark-cli commands. Feishu authorization is managed exclusively through the "Connections" panel in the Steward; the skill itself does not execute 'lark-cli auth login'. It is designed for developers who wish to operate Feishu via the command line without having to manually search for documentation. The official URL is https://api.skillhub.cn/user_d8831e99/lark-setup.
Read MoreMechanical Patent Drafting Expert: Converting Invention Technical Solutions into Preliminary Patent Applications in Chinese Standard Format
The Mechanical Patent Drafter is an Agent Skill designed for drafting initial versions of mechanical patents, identified as @user_c0efd66f/mechanical-patent-drafter. It was published by user_c0efd66f from the community, categorized as professional, and is at version 1.4.0 (requires_api_key=false). Operating in the capacity of a senior mechanical patent attorney, it takes technical solutions as input and outputs a preliminary trio: the specification abstract, the claims, and the specification. It is applicable to both invention patents and utility model patents. This article outlines its verified scope of capabilities, installation and activation methods, and typical workflows. It also notes the absence of installation commands and official examples in the scraped data, as well as inconsistencies in popularity metrics across different sources.
Read Moredsh-vscode-ide: Monaco Floating Workbench on DSH
dsh-vscode-ide is a DSH client plugin maintained by beiryang, offering a Monaco editor, independent floating windows, multi-tab support, diff, and Git and SSH/local terminals, without loading a .vsix file. Unlike the sidebar-based Monaco solution, it does not squeeze the conversation area, and is lighter than dsh-vscode-mode. When it is not yet published to npm, it can be installed by extracting the archive from GitHub Releases and running node install.mjs, after which DSH must be fully restarted. It is suitable for developers who need day-to-day editing and version control within the Harness while still wanting to preserve conversation width.
Read MoreBeelli X Community Middle Platform: Assembling Scenarios and Data Loops via Plugins in DSH
The Beelli X middle-platform plugin (dsh-beellix-plugin 0.1.1, MIT) provides plugin management, model configuration, official website catalog "Discovery", and scenario navigation within DSH. It assembles foundation, scenario, and tool plugins based on the Harness execution base and data contracts. During installation, enter the GitHub repository URL in the plugin interface. Once enabled, configure the data catalog and models via "More → Beelli X Middle-Platform". Requires Node ≥22 and DSH 0.2.1-rc.2 up to <0.3.0. Suitable for DSH developers requiring multi-scenario collaboration and local data closed-loop; please review the source code and permission boundaries before use.
Read MoreRetrieval and Analysis of Listed Company Announcement Cases: A-Share Announcement Case Retrieval and Word Report Generation
Retrieval and Analysis of Announcement Cases of Listed Companies is a community skill on SkillHub (source: community, category: professional, publisher: user_dd406368, version: 1.1.1, requires_api_key=false). It primarily utilizes the full-text search API of CNINFO (Juchao Information), supporting three search methods: keywords, stock codes, and full-text. It also supports a manual case entry mode for searching announcement cases of A-share listed companies, compiling case summaries, and generating Word reports of case analysis with links to the original texts. The article introduces the skill's positioning, core functions, trigger-word-based activation methods (trigger words include announcement cases, case analysis, regulatory cases, penalty cases, and CNINFO full-text), typical workflows, applicable scenarios, and limitations. This skill is suitable for developers and researchers engaged in research on regulatory penalties and the compilation of information disclosure cases.
Read MoreTavily Search: Integrating Tavily Retrieval into an Agent’s Search Skill
Tavily Search is an Agent search skill that conforms to the universal SKILL.md format, with coordinates @org-eyw2ohcx/tavily-ai-skills-tavily-search, version 1.0.0, publisher RealQ, source enterprise, category knowledge-management. The tag requires_api_key=true indicates that users must supply their own Tavily API Key before using it. The SkillHub page shows a rating of 100,000, 1,295 downloads, and 0 installations. This article only presents information verifiable from the provided sources; since no official, verifiable installation commands, configuration fields, or usage examples have been released, no speculation is made. Users should visit the official address to confirm details.
Read MoreAdvanced Development Journal Generator: Organizing Daily Work Reports from Conversations and Git Commits
The Senior R&D Daily Report Generator (@user_6837aee4/dev-daily-report) is a community-sourced Agent Skill categorized under office-efficiency, currently at version 1.0.0, and requires no API Key. It addresses the pain points of R&D professionals who find writing daily reports time-consuming and prone to omissions. When users mention daily work reports, daily reports, today's work, or "daily report," or wish to summarize today's development work or generate markdown reports, the Agent extracts technical tasks from conversation history, summarizes git commits, and outputs a professional markdown daily report. The article introduces its three core capabilities, activation methods based on SKILL.md, configuration without API Key requirements, typical usage examples, and usage limitations such as relying on conversation and commit integrity, summarizing without planning, and provides the official URL.
Read Moredsh-ssh-helper: SSH Connection Pool and Security Guardrails for Personal Server Operations
dsh-ssh-helper is a DSH admin-security plugin maintained by bcdws. It provides a single reused SSH connection per host, tools such as ssh_run/read/grep, optional log bypass summaries, and safety guardrails that strictly block 6 categories of system-destructive commands. It is key-only and does not store private keys. Installation requires cloning the repository, running `npm install`, and then executing `dsh plugin add link`. It is ideal for using AI to operate personal servers within DSH while controlling context and security boundaries. Licensed under MIT, version 0.5.0, it depends on DSH >=0.2.1-rc.2.
Read MoreVideo Script Generation Workflow: Batch Generation of Short-form and Long-form Video Scripts
**Preface** When producing content for products, a common pain point is that script production capacity cannot keep pace with topic selection. Short videos require brevity and speed, long videos demand complete structural integrity, and series-based columns often require the simultaneous rollout of over a dozen topics. The bottleneck in manual scriptwriting lies not in creativity, but in the repetitive labor of breaking down ideas into structures, dialogue, and durations. Introduced below is a Skill designed for such requirements: the Video Script Generation Workflow. It is contributed by community publisher **user_c5d41a7d**, version 1.0.0, categorized under **content-creation**. **What is this?** **One-sentence positioning:** A general-purpose Skill in SKILL.md format designed to generate scripts for both short and long videos, with support for the batch generation of series topics. **Origin and Maintenance:** Sourced from the community; published by us...
Read MoreDaily Tech/AI/Internet Trend Radar: Condensing 24–72-Hour Industry Updates into Actionable Content Strategies
This is a Skill maintained by community publisher user_c755f717, version 1.0.0, category data-analysis, source community, and tagged requires_api_key=false. It performs daily online searches for the latest 24 to 72 hours of developments in technology, AI, and the internet, cross-verifies sources, deduplicates, and assesses importance. It outputs hot topics, potential high-impact story angles, undercurrents on social media, items to keep tracking, and noise-reduction/editing judgments. It is suitable for content teams that need to monitor industry updates daily and turn trending topics into story angles. However, importance judgments still require human review, and the page shows 0 installs, indicating insufficient community validation; a small-scale trial is recommended. Official materials do not include installation commands or reproducible examples, and this article makes no speculative assumptions.
Read More