SenseNova-Skills: A Composable AI Office Skills Library

SenseNova-Skills is an open-source office skills library from SenseTime for agent runtimes. It organizes infographic creation, presentations, Excel analysis and research into reusable modules, helping turn source material into analysis and visual reports.

How the skills work

Each skill has its own directory and SKILL.md instructions describing its method, inputs and steps. Compatible environments include OpenClaw and Hermes Agent. An agent can select a module for one task or combine modules into a larger workflow.

Core capabilities

• Infographics: Organize reports, data and concepts visually with layouts, styles and visual review.
• Presentations: Plan pages from a topic. A standard workflow uses HTML pages, visual review and PPTX export; a creative workflow can generate images page by page.
• Excel analysis: Read and clean tables, filter records and aggregate sheets. Parquet or streaming approaches reduce memory pressure, with image-table recognition and chart-data extraction.
• Research: Structure planning, focused evidence gathering, cross-checking and synthesis. Address conflicting views or statistical definitions and continue interrupted work.

Academic, developer and social search can supply evidence. More sources alone do not establish correctness; analysis requires clear field definitions and attention to missing values.

Setup and use

1. Read README, INSTALL.md and relevant skill instructions.
2. Install in a compatible environment and configure documented model interfaces, dependencies and search services.
3. Provide material, audience, output format and scope.
4. Test a standalone task before combining modules.
5. Review figures, charts, references and readability.

A possible chain is spreadsheet analysis, research into findings, then a presentation. Pass explicit data and conclusions between modules so evidence is retained.

Applications and limits

Uses include business reports, project presentations, training and research. Visual review and streaming support the process but do not guarantee error-free or publication-ready output.

Open-source skills do not make every model or service free. Versions, dependencies, allowances and fees follow current documentation and your account. See Raccoon for a hosted experience or the official repository for local integration.