SenseNova-Skills

What is SenseNova-Skills?

SenseNova-Skills is an open-source, modular AI office skills library from SenseTime, designed specifically for agent runtime. It breaks down four core office capabilities—infographic generation, PPT creation, Excel data analysis, and in-depth research—into independent, reusable skill modules. Following the agentskills.io specification, it can be directly integrated into mainstream agent frameworks such as OpenClaw and Hermes Agent . Underlying it is the collaborative work of SenseNova 6.7 Flash-Lite and the U1 multimodal model, utilizing VLM quality closed-loop and streaming large file processing technology to achieve the capability from “providing an analysis approach” to “automatically completing the workflow and directly delivering the final product.”

Main functions of SenseNova-Skills

• Infographic generation : Supports 87 layouts and 66 styles. Through VLM automatic quality checks and multiple iterations, it transforms complex data, long reports, or business insights into high-density visual infographics.
• PPT generation : Offers standard mode (HTML pagination + VLM review → PPTX export) and creative mode (generates 16:9 U1 full-page PNG image per page), and supports automatic task distribution after parsing PDF/DOCX/MD/TXT and other formats.
• Excel data analysis : Supports multi-table reading, automatic conversion of large files (≥10,000 rows) to Parquet, data cleaning, conditional filtering and cross-table aggregation; large tables with hundreds of thousands of rows use streaming processing to avoid memory overflow; supports image and table OCR recognition and chart data extraction.
• In-depth research : Following a standardized path of “planning → multi-dimensional targeted evidence collection → cross-verification → comprehensive judgment”, it automatically identifies discrepancies in statistical standards and contradictions in industry viewpoints, and outputs visual reports with rich charts and traceable sources, supporting resume from where it left off.
• Multi-source search : Covering vertical search capabilities for academic (ArXiv, PubMed), developer (GitHub, HuggingFace), Chinese social media (Bilibili, Zhihu, Douyin) and English social media (Reddit, X, YouTube).
• Modular integration : Each skill is independently packaged as a directory+ SKILL.md, which can be called independently or combined into complex end-to-end workflows, and directly connected to Agent frameworks such as OpenClaw and Hermes Agent.

Project address for SenseNova-Skills

• Project address : https://github.com/OpenSenseNova/SenseNova-Skills

How to use SenseNova-Skills

• Automatic Agent Installation (Recommended) : Send natural language commands to the deployed Agent (such as OpenClaw, Hermes Agent), such as “Please install SenseNova-Skills from https://github.com/OpenSenseNova/SenseNova-Skills to the skills directory”. The Agent will automatically clone the repository and complete the configuration. After installation, you may need to manually restart the Agent service.
• Manual installation : By git clone https://github.com/OpenSenseNova/SenseNova-Skills.git --depth=1cloning the repository, skills/copy the subfolders under the directory to the corresponding Agent’s skill directory (such as OpenClaw’s ~/.openclaw/skills/or Hermes’ ~/.hermes/skills/), and INSTALL.mdconfigure the API Key and dependencies accordingly.
• Office Raccoon (Zero-Configuration Experience) : Directly access the Office Raccoon official website to use the intelligent office assistant. No installation, environment configuration, or API Key is required. Experience the full suite of SenseNova-Skills office capabilities for free.
• Configure API Key : After manual installation, you need to .envconfigure the SenseNova Open Platform API Key in the file (apply for a free quota at https://platform.sensenova.cn/console/keys). Some search skills also require configuring separate API Keys for each platform.
• End-to-end workflow combination : Individual skills can be used independently or combined in series as needed. For example, data analysis results can be input into in-depth research, and the research conclusions can be handed over to the PPT generation skill to form a complete workflow of “data → insight → report”.

SenseNova-Skills’ core advantages

• Modular and composable : Each skill is independently encapsulated as a directory+ SKILL.md, following the agentskills.io specification, which can be called independently or flexibly linked into complex end-to-end workflows.
• End-to-end delivery of finished products : It’s not just about outputting analytical recommendations, but about automatically completing the entire process of data cleaning, chart generation, report writing, and PPT synthesis, directly delivering a finished product that can be used in meetings and for presentations.
• VLM Quality Closed Loop : Visual language model is introduced into the generation process of infographics and PPTs for automatic quality inspection and multiple rounds of iteration to ensure that the layout, color scheme and text-image matching meet the publication-grade standards.
• Large file streaming processing : For large Excel files (≥10,000 rows), streaming reading + automatic conversion to Parquet technology is used. Even data tables with hundreds of thousands of rows can be processed stably, avoiding memory overflow.
• Native multimodal architecture : The underlying layer relies on SenseNova 6.7 Flash-Lite and the U1 model to work together—U1 is responsible for image generation, while 6.7 is responsible for planning, expansion, quality inspection and text generation, to achieve deep collaboration between images and text.
• Standardized in-depth research path : Following the methodology of “planning → multi-dimensional targeted evidence collection → cross-verification → comprehensive judgment”, it automatically identifies statistical discrepancies and contradictions in viewpoints, and supports resuming research even after interruptions.

Comparison of SenseNova-Skills with similar competing products

Application scenarios of SenseNova-Skills

• Enterprise performance and HR data analysis : Batch processing of performance appraisal Excel spreadsheets spanning multiple months and positions, automatically cleaning tens of thousands of data entries, and generating complete Word reports and HTML visualization pages that include trend analysis, job comparison, employee profiles, and improvement suggestions, which can be directly used for management meeting decisions.
• In-depth industry research and investment analysis : A single sentence can define your research needs, automatically completing multi-source searches, cross-verification of evidence, and analysis of statistical standards. It outputs a professional analysis report with more than ten charts and is traceable, suitable for investors and industry researchers to quickly understand the current market situation.
• Business presentation automation : Based on theme, audience, and style requirements, automatically plan the outline, allocate materials for each page, generate semantic HTML, and synthesize PPTX. Suitable for long text presentation scenarios such as product launches, project reports, and training materials.
• Complex information visualization : Compress long reports, business insights, or technical concepts into high-density structured infographics, supporting 87 layouts and 66 styles, suitable for marketing materials, knowledge popularization, data journalism, resume visual design, and other scenarios.
• End-to-end data research chain : Starting with raw CSV/Excel data, we first clean and analyze trends, then trigger in-depth research on outlier data (such as “reasons for rising memory prices”), and finally synthesize data conclusions and research insights into professional PPT slides, achieving full automation of the “data → insight → report” chain.
• Academic and competitor intelligence gathering : Covers multiple vertical search sources such as ArXiv, PubMed, GitHub, HuggingFace, Zhihu, and Reddit, automatically integrating fragmented information into structured analysis conclusions, suitable for the daily intelligence work of researchers, product managers, and strategic analysis teams.