dsh-token-audit: Auditing Token Spending from Local Session Logs and Presenting Actionable Cost-Saving Solutions
dsh-token-audit (lengrufeng2008) reads local ~/.dsh/sessions logs and calculates costs transaction-by-transaction based on DeepSeek's official peak/off-peak pricing. It outputs cost breakdowns, context distribution, and Savings plans. Supporting npx/global CLI, JSON comparison, and Lean presets, it can also register the read-only tool token_audit via `dsh plugin --profile web add dsh-token-audit`. The entire process is local with no data uploads; results are estimates, not invoices.
Read MoreInspection Instruction (SIP) Generation Skill: From Product Technical Documents to Standardized SIP
Standardized Inspection Instruction (SIP) generation skill, published by EngineerHub, version 1.1.0, categorized as professional and does not require an API Key. It parses product technical documents in multiple formats, extracts key information such as inspection items, technical requirements, and acceptance criteria, and organizes them into standardized inspection instructions according to user-provided Excel or Word templates, suitable for scenarios where quality inspection standards or supplier inspection requirements need to be generated. This article describes its positioning, core features, installation access, typical usage, and applicable precautions; page data (rating 100000, downloads 920) are from SkillHub capture results.
Read MoreMiddle School Classic Literature Coaching: Transforming Close Reading Assignments into an Iterative Process
"Middle School Classic Literature Tutoring" is an education-category Skill from the SkillHub community (`@user_2c769e18/classic-reading-tutor`, version 1.0.0, published by user_2c769e18). Based on classic literature and school reading/guidance manuals, it generates themed close-reading assignments—such as excerpts and summaries, comprehension and reflections, and mind maps (output as Word documents)—for one or multiple middle school students. It also provides structured, teacher-perspective feedback on generated assignments to drive revision and iteration. Additionally, it supports classic literature reading framework analysis, multi-style student role modeling, and the construction of a verifiable database of expert literary criticism. No API key required. It is not applicable to university-level academic research, non-literature textbook tutoring, pure content summarization, or general essay grading unrelated to reading manuals. During verification for this article, the official webpage scraping failed, and installation commands as well as variations across different clients could not be verified, which has been transparently noted in the main text.
Read Moredsh-tdd-guard: Pre-commit Test Gate in DSH
dsh-tdd-guard is a DSH workflow plugin maintained by lengrufeng2008. It runs `npm test` on the project before committing and allows for the installation of a pre-commit hook tagged with `dsh-tdd-guard`. It provides `status`, `check`, `install`, and `uninstall` commands, and `tdd_check` is available within DSH. It acts as a test-before-commit gate rather than enforcing test-first during the editing phase. Node >=20, MIT 0.1.0.
Read MoreZhenPiaoCloud Smart Invoicing: Conversational Digital Invoice Issuance Skill
This article introduces the community Skill "Zhenpiao Cloud Intelligent Invoicing" (`@user_e5cefaa2/issue-an-invoice`, v1.0.7, business-ops, requires_api_key=true) developed by Shuzu Technology. It integrates enterprise invoicing workflows into a conversational interface: utilizing natural language to parse invoicing elements and generate to-do records, verifying the seller and completing buyer information via Zhenpiao Cloud enterprise master data, AI-matching commodity tax classification codes while validating compliance regarding goods, tax rates, and amounts, and organizing information for user confirmation before calling the Zhenpiao Cloud MCP service to issue blue digital electronic invoices. It also supports conversation context preservation, historical invoice retrieval, and red-letter invoice reversal. The article outlines core features, installation and activation methods, typical conversational use cases, and important notes such as API Key requirements, community sourcing, and the necessity for testing and validation.
Read MoreFiscal & Tax Advisor: Q&A on Finance and Taxation in Real Estate and Commercial Property
"Financial & Tax Advisor" is a community Skill published on SkillHub by user_cd2cd819 (Location: @user_cd2cd819/fin4u, Version: 1.0.0, Category: Professional). It is designed to encapsulate the expertise of a dual CPA and CTA certificate holder with 15 years of practical financial and tax experience in real estate and commercial properties for Agent invocation. Its capabilities cover tax planning and audit response, comprehensive budgeting and fund control, investment analysis and financial modeling, as well as business analysis and audit compliance. It serves sectors including real estate development, industrial parks, commercial properties (office buildings + hotels), the elderly care industry, and the education industry. The Skill includes specific trigger keywords for Land Appreciation Tax liquidation, the Business Tax to VAT reform, Real Estate Tax, Corporate Income Tax, cross-border taxation, and High-Talent Pass tax matters. While no API Key is required, the retrieved data lacks official installation commands and distinctions for Cursor, Codex CLI, or Claude Code; therefore, installation steps should be based on the SkillHub page. The page displays a rating of 100,000 and 912 downloads, yet shows 0 installs, indicating very few actual users. It is recommended that tax-related conclusions be reviewed by qualified personnel before use.
Read Moredsh-skill-scan: Static Read-Only Initial Security Screening for DSH Skills and Plugin Packages
`dsh-skill-scan` (lengrufeng2008, admin-security) performs static, read-only scans of DSH third-party skills and plugin bundles without executing the bundles. Heuristics cover instruction overrides, `curl|sh`, credential paths, `eval`, obfuscated Unicode, npm lifecycle scripts, and Cordis `!!js`. Supports both CLI and DSH plugin interfaces; `PASS` indicates non-security proof, while `BLOCK` requires manual review with line numbers. Node >=20, MIT 0.1.0.
Read MoreDuan Yongping Perspective Skill: Integrating Six Value Investing Mental Models into an Agent
This article introduces the "Duan Yongping Perspective" Skill on SkillHub / ClawHub (@sjj2026/duanyongping-perspective, v1.0.0, MIT-0, publisher: sjj2026). It distills the mindset from over two decades of Duan Yongping's public statements into six mental models (buying stocks is buying companies, the principle of the circle of competence, margin of safety and contrarian investment, long-term holding and simplified portfolios, the "Ben Fen" philosophy, and "daring to be a follower") and an expression DNA. By utilizing a standard SKILL.md, it allows Agents to respond to investment research questions with Duan Yongping-style logic and tone. The article covers the three-stage response workflow (Question Classification → Duan Yongping-style Research → Duan Yongping-style Answers), four trigger words (Duan Yongping, Da Dao, Ben Fen, Value Investment), the `openclaw skills install` command, reproducible usage examples, and its applicable scenarios and limitations: the methodology originates from the 2010s, it is unsuitable for short-term trading or technical analysis, there are controversies surrounding portfolio concentration, and trigger words are primarily in Chinese. The official URL is clawhub.ai/sjj2026/duanyongping-perspective.
Read MorePrimary School Oral English Course Designer: Developing the Entire Course into a Deliverable Plan
The Elementary Oral English Course Designer is an education-category Skill within the SkillHub community (v1.0.0, no API Key required). It is designed to create complete oral English course plans ranging from 4 to 40 sessions for elementary school students in mainland China. It covers student profiling, entry-level diagnostics, course positioning, capability objectives, thematic units, session sequencing, cyclical review, classroom tasks, stage-based deliverables, oral assessment, parental evidence, and instructional delivery requirements. It is suitable for English training institutions, teams of local and foreign teachers, and independent teachers to design oral courses or review the progressive difficulty and observability of outcomes in existing curricula; it is not a single lesson plan or a tool for marketing/enrollment.
Read Moredsh-skill-library: Install Once, 100 Executable Checklist Skills
dsh-skill-library is a DSH workflow plugin maintained by lengrufeng2008. Within a single installable bundle, it contains one skill provider and 100 structured checklist skills, covering coding, testing, security, data, DevOps, product, writing, research, career, marketing, and operations. Each skill includes whenToUse, methodology, acceptance criteria, and boundaries. Skills can be invoked by name or discovered by the router from the catalog. Installation follows the local path examples in the README. Note that skills are checklists rather than correctness proofs; spot-check is recommended before use. Licensed under MIT. Requires Node >=20. Version 0.1.2.
Read MoreJLC Online Aluminum Alloy Agent Skill: Delegating Enclosure Modeling Attributes to AI
The JLC online aluminum alloy agent skill (@user_6b95f8df/jlcfa-aluminum-modeling, version 1.0.4, sourced from the community and categorized under dev-programming) is designed for the 3D modeling of JLC FA aluminum alloy enclosures. It is responsible for building and reviewing modeling properties, covering cutouts, threaded holes, countersinks, anodizing, UV printing, laser marking, panel machining, and enclosure machining. Tagged with requires_api_key=false, it does not require an API key. Based on the official introduction, this article outlines its positioning and capabilities, and notes that the provided materials lack official installation commands, configuration fields, and reproducible usage examples. Therefore, users should refer to the official page before use. Additionally, it provides applicable scenarios and precautions to consider.
Read MoreMaster of Poetry Creation: Standardizing Poetry Writing into a Reusable Workflow
"Poetry Master Plus" (@user_60a201c0/poetry-master-plus), published by user_60a201c0, is a community-sourced content creation tool (Category: content-creation, Version: 1.3.4) that requires no API key. It operates on the first principles of poetry defined as "Imagery as the soul, sincerity as the backbone, and blank space as the rhythm." It claims proficiency in both classical metrics and modern verse, aiming to produce poems with substance that are suitable for recitation. This article outlines its positioning and capability boundaries. It also notes that installation commands, configuration fields, and official examples are missing from the retrieved materials, so users should verify this information at the official source before use. Ratings, download, and installation data are sourced from a SkillHub page crawl snapshot; please confirm the statistical methodology before citing.
Read Moredsh-secret-guard: DSH Local Key Scanning and Masking
dsh-secret-guard is a DSH admin-security plugin maintained by lengrufeng2008. It performs local credential pattern scanning and redaction on prompts, logs, diffs, and files. Matched content is returned in a "redacted" format, ensuring complete secrets are not transmitted back. It supports rules for GitHub/AWS/OpenAI-style keys, private key blocks, and JWTs. The plugin offers CLI usage as well as `secret_scan` and `secret_redact` conversational commands, with no network access or telemetry. Based on pattern matching, no matches do not guarantee security, and redaction may break syntax requiring review. Requires Node >= 20. MIT License. Installation identifier: lengrufeng2008/dsh-secret-guard. For details, please refer to the SkillHub catalog and GitHub README.
Read MoreClassic Endorsement Arsenal: Turning Your WeRead Bookshelf into Content Proof
"Classic Endorsement Arsenal" is a Skill designed for content creators to systematically transform books from their WeRead (WeChat Reading) shelves into classic endorsement material for short videos, livestreams, paid columns, and blogs. It automatically fetches book information and popular highlights (including the number of users who marked them), generating clean Markdown notes in the user's personal note repository, with optional synchronization to Dedao Brain and IMA Knowledge Base. This Skill is not bound to a specific persona; a first-time use wizard collects the creator's perspective, ensuring all outputs are generated from the user's viewpoint. The ultimate goal is content endorsement, not simple book summaries. Target users include self-media creators, knowledge bloggers, lecturers, consultants, and writers. Trigger phrases cover: "Turn a WeRead book into endorsement material," "Weave books from my shelf into an arsenal," "Sync back to Dedao Brain/IMA," and "How to use books in my content."
Read Moredsh-science-legion: Layered Multi-agent Science Campaign and Verifiable Claims
dsh-science-legion is a DSH workflow plugin (v0.2.0, MIT) maintained by lengrufeng2008. It organizes tasks related to mathematics, physics, astronomy, engineering, and biomedical evidence through science-legion routing and legion-command directives. The default configuration includes six agent teams featuring an independent verifier and a red team, emphasizing verified claims and token budgets; it is not an open-ended problem solver. Installation: `dsh plugin --profile web add github:lengrufeng2008/dsh-science-legion`. Usage example: `Use science-legion on: prove or refute ...`. For large-scale tiers, use `estimate-campaign` to plan costs and set limits in the BRIEF first.
Read MorePomegranate Daily Hot Topics: Converting Today's Trends into an Actionable Topic List
"Pomegranate Daily Trending Topics" is an Agent Skill designed for content creators, published on SkillHub at @user_7f6c97a1/shiliu-daily-trending (Version 1.0.0). Classified under content-creation and sourced from the community, it is published by user_7f6c97a1. This Skill searches the web daily for trending topics across four domains: technology, finance, business, and society. It performs a cross-scoring based on three elements: corporate dynamics, technological breakthroughs, and user-perceived value (scoring 0-2 for each). Topics with a total score ≥4 are selected, with a perfect score designated as a "premium topic." The output is a ranked list of topics, each accompanied by suggested angles, user value points, recommended content formats, and warnings. The official page displays a rating of 100000, 966 downloads, and 0 installations. The Skill is tagged with requires_api_key=false, indicating no additional API key is needed. Official documentation does not provide a one-line installation command or specific installation instructions for Cursor, Codex CLI, or Claude Code, but offers trigger phrases such as "What are the trending topics today?", "Show me today's tech and finance highlights," or "What can I film today?". The article covers core features, installation access, typical usage, applicable scenarios, and precautions. It is suitable for creators needing a stable daily output of topics. Note that it only assists in topic selection and angle planning, not the actual writing. It requires an internet connection, and users must independently verify facts and assess compliance risks regarding the output.
Read MoreValuation Modeling Expert: Running DCF / PE / PB / WACC Valuation Processes in Agent
This article introduces the community skill "Valuation Modeling Expert" (ID: @u_8ee34d77/valuation-model-free, version v1.0.0, category: professional, published by u_8ee34d77) on SkillHub. Designed for financial analysis scenarios, this skill claims to cover the full workflow of DCF/PE/PB/WACC, with functional tags "valuation" and "model". It supports Chinese interactions, structured outputs, and error handling, making it suitable for independent developers, enterprise teams, and automated workflows. The skill explicitly states it is not intended for complex requirements beyond its capabilities. The tag requires_api_key=true indicates that API key configuration is required for invocation. The retrieved materials did not provide official installation commands, platform-specific integration differences, or details on the implementation and parameters of the four valuation methods. Therefore, the main text only includes verifiable information about coordinates, version, source, and classification. Installation steps should follow the official page, and it is recommended to perform baseline validation with known-answer samples before deployment. The page shows a rating of 100,000 and 907 downloads (as of the retrieval time).
Read MorePDF Documents, Intelligent Extraction, and Format Conversion: Turning PDFs into Structured Markdown
Introducing the SkillHub community skill "Intelligent PDF Document Extraction and Format Conversion" (@user_a10ab3ad/markdown-themeable-pdf, version 1.0.1, categorized under office-efficiency). This skill efficiently converts PDFs into structured Markdown, preserving key information and supporting customizable output formatting without requiring an API Key. The article details its core functionalities, installation and activation methods, typical usage, applicable scenarios, and limitations, along with providing the official URL.
Read Moredsh-sbom: Generating CycloneDX 1.5 SBOMs for DSH Plugin Packages
dsh-sbom is a DSH plugin maintained by lengrufeng2008 (MIT, v0.1.0, Node≥20). It generates CycloneDX 1.5 SBOMs from the plugin package and installed top-level node_modules, and supports both the CLI and sbom_export. It operates offline, without a transitive dependency graph or vulnerability assertions, making it suitable for manifest archiving rather than complete supply chain proof. Installation can be requested from DSH via the SkillHub identifier lengrufeng2008/dsh-sbom; review the source code and process permissions before use.
Read Moredsh-repomix: Capped Read-Only Repository Bundling for DSH Context Handover
dsh-repomix is a DSH workflow plugin maintained by lengrufeng2008 that packages bounded, read-only repositories into a single XML or Markdown file to facilitate context handoff. It supports per-file and total byte limits; when truncation occurs, it marks the content as "truncated" rather than silently omitting it. By default, it skips .git, node_modules, build artifacts, and common credential paths, though it is not a substitute for security auditing. You can install lengrufeng2008/dsh-repomix to DSH via SkillHub, or invoke it using `node bin/dsh-repomix.mjs` and the `repo_pack` tool. Requires Node >= 20. MIT License. v0.1.0.
Read More**WeCom To-Do: Read and Write WeCom To-Dos in Agent**
This article introduces the SkillHub community skill "WeCom Todo," with coordinates @user_d8831e99/wecomcli-todo. It is version 1.0.0, categorized under office-efficiency, sourced from the community, and published by user_d8831e99. The page is labeled with requires_api_key=false, a rating of 100,000, 912 downloads, and 0 installations. The skill targets Agent scenarios requiring read and write operations on WeCom (Enterprise WeChat) todos, covering querying the todo list, retrieving todo details, creating todos, updating todos, deleting todos, changing user processing status, and assignment-related expressions listed in the introduction. The article explains capabilities item by item and organizes the colloquial phrasing from the skill introduction into a trigger scenario mapping table to facilitate intent routing after integration. Since official materials do not provide installation commands or configuration fields, the article only records the SkillHub page and official address, avoiding speculation on installation methods; differences in enabling the skill across Cursor, Codex CLI, and Claude Code are omitted due to the lack of verifiable content. regarding applicability, the article reminds users that since the materials do not specify permission scopes, field constraints, error codes, or rate limit information, target todos should be confirmed before performing write operations such as delete, accept, or reject. As the source is the community, it is recommended to return to the SkillHub page to verify the source and the latest version before formal integration.
Read MoreCross-border E-commerce Standalone Site Facebook Product Testing Agent: A Three-Agent Workflow for Regional Market Analysis, Facebook Ad Library Product Selection, and Creative Generation
This Skill is a community-sourced project on SkillHub with the internal identifier `independent-store-agent-suite`, version 1.0.0. Published by `user_96df57c8`, it is categorized under `business-ops`. At the time of data scraping, it had a rating of 100,000, 952 downloads, and 0 installations. It is tagged with `requires_api_key=false`, meaning no additional API Key is required for use. The Skill deconstructs the product selection and creative material workflow for cross-border independent websites into three Agents that execute in a fixed sequence: 1. **Agent 1** performs regional market analysis, outputting 10–30 product directions and 30–60 English search keywords for the Facebook Ads Library. 2. **Agent 2** must ask for the desired number of candidates to collect. It then searches the Ads Library with settings set to `Region = All` and `Ad category = All ads`. It records specific ad ID links, advertisers, landing pages, and video download status. Candidates are scored based on sellability, creative strength, copy strength, supply chain feasibility, and risk score. Agent 2 outputs a CSV of candidates and pauses for user confirmation. 3. **Agent 3** processes only `approved` products. It first verifies if the ad, video, and landing page refer to the same product, then creates a `product_identity.json`. It generates a fixed output of 2 main images plus 5 detail images, with independent prompts for each. Detail images utilize the language of the target market. Generated images must be displayed to the user for review before proceeding. The fixed sequence is: **Agent 1 -> Agent 2 -> User Confirmation -> Agent 3 -> User Review -> Packaging after Approval**. Skipping Agent 1 or Agent 2 is not allowed. **Limitations:** * The Facebook Ads Library does not provide data such as clicks, traffic, conversions, or CPA; the Skill does not fabricate this data. * Without browser or download capabilities, Agent 2 enters manual guidance mode: videos that cannot be downloaded are marked `needs_manual_download`; inaccessible landing pages are marked `manual_source_required`; if image generation is unavailable, it outputs `prompts_only`. * The Skill does not process high-risk products related to knives, firearms, food, drugs, health supplements, medical efficacy, finance, politics, or religion. * No installation commands (npm, pip, or git clone) are provided in the documentation; installation instructions should follow the official page.
Read MoreZhixia Selection Data - Shopee Big Data Analysis: Enabling Agents to Read Shopee Selection Data Directly
**Introduction to ZhiXia Product Selection Data - Shopee Big Data Analysis Skill:** This Skill analyzes sites, categories, products, shops, brands, trends, and hot search keywords via Shopee MCP. It is applicable for market size assessment, category product selection, competitor research, ranking analysis, shop/brand diagnosis, and keyword trend analysis. Published by Guangzhou Mengla Information Technology Co., Ltd. (Enterprise Source), it is categorized as data-analysis, Version 1.0.5, with the tag requires_api_key=true, indicating an API Key is required. As the currently scraped data does not provide installation commands, configuration fields, or enablement steps for various editors, and the official page failed to load, this article truthfully states that these installation and enablement details cannot be verified; it is suggested to refer to the official address. Meanwhile, precautions regarding data coverage, update frequency, authorization scope, and version changes are also noted.
Read Moredsh-repo-map: Generating compact directed repository graphs with regex
dsh-repo-map (lengrufeng2008) is a "memory"-category plugin of the DeepSeek Harness community. It scans repositories using regex heuristics to generate a compact map of source files and top-level symbols, without relying on LSP or call graphs. This article introduces its positioning, the division of responsibilities with dsh-code-intel / dsh-repomix, the SkillHub installation method, CLI and Harness usage, and its capability boundaries. Version 0.1.0, MIT licensed, requires Node 20+.
Read MoreInstant Thesis Search: Local Academic Paper Retrieval and Export
Paper Quick Search is a community skill on SkillHub, located at @user_3cdbf8ea/paper-search. Version 1.0.4, published by user_3cdbf8ea, it implements local retrieval and export of academic papers using pure Python and includes a built-in index of classic papers. It is tagged with requires_api_key=false, indicating that no additional API key is required. The article verified page information such as name, coordinates, version, source, category, rating, and download count, and noted that installation commands and typical usage examples were missing from the retrieved materials; no content was fabricated.
Read More