SenseNova-Skills is an MIT-licensed JavaScript repository of modular Agent Skills that extends the SenseNova model family with end-to-end office and productivity capabilities, built for developers and teams assembling AI-powered office assistants on Agent Skills-compatible runtimes.

What it is

SenseNova-Skills is a collection of office-automation skills from the OpenSenseNova project, written in JavaScript and released under the MIT licence. Each skill lives in its own directory and declares its triggers, capabilities, and execution flow in a SKILL.md file, following the Agent Skills convention defined at agentskills.io. The repository sits in the AI & Machine Learning / AI Interaction & Interfaces category and plugs SenseNova models into agent runtimes such as OpenClaw and hermes-agent. The suite covers image generation and visualization, slide-deck (PPT) generation, Excel data analysis, deep research, HTML experiences, team collaboration, and proactive project tracking, and the skills may be used standalone or composed into end-to-end workflows.

The problem it solves is the glue layer between a capable model and concrete office output. Instead of writing bespoke prompts and per-task integration code for each document format, a user copies skill directories into an agent's skills folder and the runtime picks them up, with the model supplying reasoning and multimodal understanding. That makes the unit of reuse a portable skill directory rather than a one-off prompt, letting slide decks, spreadsheets, research briefs, and infographics share one execution convention. The repository also documents both the international and mainland China SenseNova API flows.

Key capabilities

• Each skill is a directory carrying a SKILL.md manifest that declares triggers, capabilities, and execution flow, following the Agent Skills convention.
• Image generation and visualization, documented in docs/sn-infographic-examples.md with nearly 100 generation cases and their prompt designs.
• Slide-deck (PPT) generation.
• Excel data analysis.
• Deep research.
• HTML experiences.
• Team collaboration and proactive project tracking.

Who uses it and how

• Teams that would rather not provision environments, API keys, and runtimes can use the same skills bundled into Raccoon, the hosted option pairing SenseNova models with the full Cowork-Skill suite, enterprise-grade security, and a zero-setup flow.
• Developers running OpenClaw copy skills into ~/.openclaw/skills/, while those on hermes-agent use ~/.hermes/skills/; the agent service may need a manual restart before new skills are picked up.
• Organizations working across regions choose the matching API, pointing at https://token.sensenova.ai/v1 internationally and https://token.sensenova.cn/v1 in mainland China, with the documentation page, API key, base URL, and model name all drawn from the same region.
• Raccoon reports use by 15M+ individual users and thousands of enterprise customers, and ships a local Agent OS positioned next to a user's files, work context, and personal habits.

Getting started

Install by cloning the repository and copying its skill subdirectories into the agent's skills directory, as in "git clone https://github.com/OpenSenseNova/SenseNova-Skills.git --depth=1" followed by "cp -r SenseNova-Skills/skills/* ~/.openclaw/skills/", with the full walkthrough in INSTALL.md. Alternatively, hand the repository URL to the agent and let it install the skills, or use the hosted Raccoon option at xiaohuanxiong.com for a zero-setup experience.

How it compares

SenseNova-Skills is not framed against a named roster of paid competitors; it sits inside the agent runtimes that host it, with OpenClaw and hermes-agent providing the runtime that consumes the SKILL.md skills. Raccoon is the managed alternative from the same project, bundling the same Cowork-Skill suite with SenseNova models and enterprise-grade security, so the choice is between controlling the runtime and model endpoint yourself or accepting a zero-setup hosted flow.

When to use it — and when not to

Self-hosting means operating an Agent Skills-compatible runtime such as OpenClaw or hermes-agent and supplying SenseNova Platform API credentials, and installation is a matter of copying directories and restarting the agent rather than using a package manager, container image, or compose file. Anyone who wants no environment setup, no API-key management, and no runtime maintenance should choose Raccoon instead. The sharpest practical pitfall is regional mismatch, because the documentation page, API key, base URL, and model name must all come from the same region, and the repository provides an installation walkthrough rather than a managed service.

git clone https://github.com/OpenSenseNova/SenseNova-Skills.git --depth=1
mkdir -p ~/.openclaw/skills
cp -r SenseNova-Skills/skills/* ~/.openclaw/skills/