OpenSenseNova/SenseNova-Skills

Academic Research searches academic papers, reads full texts, and traces references and citations across sources like arXiv, Semantic Scholar, PubMed, and Wikipedia. Use it for literature review, related work discovery, and deep paper reading.

China Market Search locates Chinese market, industry, macro, trade, procurement, and regulatory data from official free sources without registration or API keys. It uses a Python script for A-share disclosures and browser-based access to government databases.

Chinese Social Search searches content on Chinese social platforms including Bilibili videos, Zhihu Q&A, and Douyin videos. Use it when research requires Chinese social media data; Xiaohongshu and Weibo fall back to browser-use on public web pages.

Creative PPT generates a presentation one full-page 16:9 PNG per slide, using LLM/VLM for text and T2I image generation with web image search fallback. Use it via sn-ppt-entry to turn a topic into creative, image-forward slides.

Deep Research orchestrates multiple specialist agents to conduct systematic research, fact-checking, competitive analysis, and report generation when a request requires cross-source verification or a detailed deliverable.

Developer Search queries GitHub, Stack Overflow, Hacker News, and HuggingFace to find code examples, issues, technical Q&A, discussions, and models. Use it when you need to locate code, answers, or model info across developer platforms.

Document Analysis parses Word, PDF, and PPT files to extract text, tables, charts, and formatting, then performs cross-document aggregation and numerical checks. Use it when users ask to analyze, extract, or compare content from .docx, .pdf, or .pptx files.

Excel Data Analysis Workflow orchestrates multi-step Excel analysis: reading multi-sheet files, cleaning data, filtering and extracting subsets, aggregating across sheets, and exporting results as Excel or CSV, with automatic large-file handling via Parquet for files over 10k rows.

Financial Data Search retrieves financial market data, company fundamentals, price history, K-lines, disclosure documents, and financial news for A-shares, Hong Kong stocks, and global tickers using yfinance, mootdx, and browser automation.

Image Caption Analysis extracts text and structured data from image files such as charts, tables, screenshots, and diagrams using a vision model, then parses, visualizes, and exports the results to Excel or CSV.

Image Style Imitation generates a new image that imitates the style of a reference image while updating content based on user intent. Use it when requests ask to imitate style, keep style with new content, or perform style transfer with content changes.

Infographic generates professional infographics with various layout types and visual styles. It analyzes content, recommends layout and style, and generates publication-ready visuals for requests like 'infographic', 'visual summary', or '信息图'.

Large-Scale Excel Analysis provides streaming reads, Parquet conversion, and memory optimization for Excel files with 10,000+ rows. Use it when analyzing large datasets to avoid timeouts and out-of-memory errors.

Markdown to HTML Report turns long Markdown documents into self-contained HTML pages with editorial layout, visual design, and evidence hierarchy. Use it when asked to convert a report to a web page, share it, or improve its readability and visual design.

PPT Doctor checks the environment required by the PPT skill family, validating API keys, Node runtime, sn-image-base, and optional dependencies, then writes .env interactively. Run it before sn-ppt-entry or after changing .env to confirm setup.

PPT Entry collects presentation requirements and dispatches to AI generation modes to create slide decks. Use it when a user asks to make a PPT or presentation.

PPT Standard Pipeline generates PowerPoint decks through a deterministic staged CLI that wraps all LLM, VLM, and text-to-image calls. It supports fast and standard modes, taking task_pack.json/info_pack.json and producing PPTX or PDF.

PPT Workbench opens the AI PPT editing WebUI for existing SenseNova HTML slide decks so users can preview, inspect, or edit them without regenerating slides. Use it when opening or editing a previously generated deck in the workbench.

Prepare Citations post-processes markdown reports by deduplicating [^key] citations, converting them to [N] numbering, and appending a bibliography. It is used when a final draft needs its footnotes cleaned up for publication.

Report Format Discovery compares candidate presentation formats (research report, academic paper, analytical table, decision memo, custom) and recommends one using authoritative standards and real examples. Use it before planning research dimensions when the final deliverable format is unknown.

Research Report Templates provides structured outlines for research reports, chosen by the reader's cognitive task and domain conventions. Use it to structure reports with appropriate sections, tables, and diagrams.

Resume Image Generator creates a designed portfolio-resume image from resume content provided in conversation text, using optional style instructions and a fixed portfolio-resume layout. Use it when someone asks for a resume image, portfolio resume, 简历图, or personal resume visual design.

SN Image Base provides low-level APIs for image generation, image recognition, and text optimization for upper-layer skills in the SenseNova-Skills project. It is a base-layer skill that calls backend services without input preprocessing.

SenseNova Environment Doctor validates the SenseNova-Skills environment, checking installation, Python dependencies, and required environment variables. It prompts for missing variables, saves them to .env, and reloads the environment before other skills run.

Serper Image Search finds candidate images via Google-backed Serper.dev API, returning image URLs, page URLs, titles, and source domains. Use it when an agent needs to discover images or source pages for a query.