# SenseNova-Skills: Modular AI Office Skill Pack for OpenClaw and Hermes Agent

> SenseNova-Skills is a collection of modular skills for the SenseNova model family, designed to run inside Agent Skills-compatible runtimes such as OpenClaw and hermes-agent. The repository covers image generation, slide deck creation, Excel analysis, deep research, HTML experiences, team collaboration, and project tracking.

**OpenSenseNova/SenseNova-Skills** — Modular SenseNova skills for building AI-powered office assistants and productivity workflows

- Repository: https://github.com/OpenSenseNova/SenseNova-Skills
- Stars: 5,642 · Forks: 392
- Language: JavaScript
- License: MIT
- Published: 2026-09-10 · Updated: 2026-09-10 · Language: en
- Canonical page: https://hysenlabs.com/projects/opensensenova-sensenova-skills

## What SenseNova-Skills Is and Who It Serves

SenseNova-Skills is a repository of agent skills built for the SenseNova model family from OpenSenseNova. Each skill lives in its own directory under skills/ and declares its triggers, capabilities, and execution flow through a SKILL.md file, following the Agent Skills convention defined at agentskills.io.

The intended users are teams and developers who operate Agent Skills-compatible agent runtimes and want to add concrete, end-to-end office capabilities to those agents. The skills are designed to work with two specific runtimes: OpenClaw and hermes-agent from NousResearch. They are not standalone scripts; they need an agent runtime to schedule and dispatch them.

The repository also notes that the full skill suite is bundled into Raccoon, an office product from xiaohuanxiong.com that provides a managed environment. Teams that do not want to provision their own runtime, API keys, and environment can access the same capabilities through Raccoon instead.

## How the Skill Architecture Works

Each skill in SenseNova-Skills is a self-contained directory. A SKILL.md file inside each directory declares what the skill does and how the runtime should invoke it. This structure means skills can be added, removed, or updated independently without modifying the agent runtime itself.

The README distinguishes between skill tiers. Lower-tier skills provide foundational capabilities that higher-tier skills call. For the image stack, sn-image-base is the Tier 0 layer, exposing text-to-image generation (sn-image-generate), image editing (sn-image-edit), image recognition (sn-image-recognize), and text optimization (sn-text-optimize) through a unified sn_agent_runner.py script. Upper-tier skills such as sn-infographic are built on top of sn-image-base rather than calling the SenseNova API directly.

The sn-infographic skill documents its process: it scores prompt quality automatically, selects from 87 layout options and 66 style options, runs multiple generation rounds with VLM review, ranks the results, and outputs publication-ready infographics. It supports SenseNova U1.5 Lite with native 4K output.

Each skill category has its own full guide in the docs/ directory. For example, the image and visualization skills are documented in docs/sn-image-generate_en.md, and the presentation skills in docs/sn-ppt-generate.md.

## Installing the Skills and Configuring the Environment

The recommended path is to ask the agent runtime to install the skills itself by handing it the repository URL. The README gives this example instruction:

"Please install SenseNova-Skills from https://github.com/OpenSenseNova/SenseNova-Skills into your skills directory."

After the agent clones and copies the skills, the runtime service may need a manual restart to pick up the new skill directories. The README notes this explicitly.

For manual installation, clone the repository and copy the skill subdirectories:

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

For hermes-agent, the target directory is `~/.hermes/skills/` instead of `~/.openclaw/skills/`. The --depth=1 flag keeps the clone shallow and avoids downloading full repository history.

The .env.example file in the repository lists all the environment variables required across the skill categories. The SenseNova-specific variables include SN_API_KEY, SN_BASE_URL, SN_IMAGE_GEN_API_KEY, SN_IMAGE_GEN_BASE_URL, SN_IMAGE_GEN_MODEL, SN_CHAT_API_KEY, SN_CHAT_BASE_URL, SN_CHAT_MODEL, and others for text and vision tasks. Web search skills additionally require SERPER_API_KEY, and academic search requires DEEPXIV_TOKEN and OPENALEX_API_KEY. The full list covers Chinese platform tokens (ZHIHU_COOKIE, DOUYIN_COOKIE, BILIBILI_COOKIE), international social platforms (TIKHUB_TOKEN, YOUTUBE_API_KEY), and developer search tokens (GITHUB_TOKEN, HF_TOKEN).

The README specifically warns that the docs page, API key, base URL, and model name must all come from the same region. The international base URL is https://token.sensenova.ai/v1 and the mainland China base URL is https://token.sensenova.cn/v1.

## Skill Categories and Their Coverage

The image and visualization category contains five skills. The environment doctor skill (sn-image-doctor) validates the installed environment, checks dependencies, and interactively fills missing values into a .env file. This is useful as a first step when setting up the skill suite on a new machine. The other image skills handle text-to-image generation, image editing, image imitation from a reference, and resume image generation.

The presentations category provides PPT generation skills, documented in docs/sn-ppt-generate.md.

The Excel analysis and deep research categories provide data analysis and research workflows. The examples/ directory contains end-to-end example workflows including employee performance analysis, generative AI research, and investment memo creation (memory-price-end2end-analysis).

The team collaboration and project tracking skills extend the suite toward asynchronous workflows, covering use cases where an agent monitors a project and proactively surfaces relevant updates.

The Biweekly Report.md and Biweekly Report_CN.md files in the root document changes across release cycles.

## Limitations: SenseNova API Dependency and Regional Constraints

Every skill in this repository assumes a SenseNova model at the other end. The image generation skills reference SN_IMAGE_GEN_MODEL explicitly, and the chat skills reference SN_CHAT_MODEL. None of the skills are designed to swap in an alternative provider at runtime; the API structure, model names, and authentication all assume the SenseNova platform.

The two-region setup (international and mainland China) creates a practical constraint: teams need to know before deployment which region they are in, and they must use consistently matching credentials. The README states this is a documented failure mode, not just a theoretical concern.

The skills also require a running Agent Skills-compatible runtime. There is no standalone Python entrypoint for running a skill outside of an agent context. Teams evaluating whether SenseNova-Skills fits their needs must first have a working OpenClaw or hermes-agent installation.

The repository has no GitHub releases. The last push was on 2026-09-17, which is recent. Per-category Python dependencies are documented in each skill's full guide, not in a single requirements file, so auditing the complete dependency tree requires reading multiple documentation files.

## Comparison with Claude Tool Use

The closest conceptual alternative is using Claude's tool use capability directly, where a model calls defined tools in response to user requests. The difference in approach is architectural: Claude tool use defines tools as JSON schemas that the model calls when appropriate, and each tool is implemented by the developer. SenseNova-Skills defines skills as SKILL.md files that an agent runtime dispatches, with the SenseNova model as the reasoning engine and the skills providing the concrete action implementations.

SenseNova-Skills is more opinionated about the action layer: image generation, PPT creation, and research workflows are pre-built skills rather than tools the developer defines. That pre-built quality means faster startup for teams whose needs match the skill categories, and less flexibility for teams whose workflows do not.

## Conclusion

SenseNova-Skills is a practical option for teams already on the SenseNova platform who want structured, composable AI office capabilities without building each integration from scratch. It is not the right fit for teams using other model providers, since the skill implementations depend on SenseNova-specific API endpoints and model names. Before adopting, confirm which API region applies and obtain the corresponding keys from platform.sensenova.ai or platform.sensenova.cn, since mixing region endpoints and keys is documented as a failure mode. The sn-image-doctor skill can verify the environment before full deployment.

## FAQ

### Which agent runtimes does SenseNova-Skills support?

The README recommends OpenClaw and hermes-agent from NousResearch. For OpenClaw, skills are installed to ~/.openclaw/skills/; for hermes-agent, the target is ~/.hermes/skills/.

### What SenseNova API keys does the skill suite require?

The .env.example file lists separate keys for image generation (SN_IMAGE_GEN_API_KEY), chat (SN_CHAT_API_KEY), text (SN_TEXT_API_KEY), and vision (SN_VISION_API_KEY), along with corresponding base URLs and model names. Each region (international or mainland China) has its own credentials.

### How can I verify my SenseNova-Skills environment is set up correctly?

The sn-image-doctor skill is designed as an environment validator. The README describes it as checking the sn-image-base installation, Python dependencies, and required environment variables, and interactively filling any missing values into a .env file.

## Sources

- [Issues](https://github.com/OpenSenseNova/SenseNova-Skills/issues)
- [License: MIT](https://github.com/OpenSenseNova/SenseNova-Skills/blob/main/LICENSE)
- [OpenSenseNova/SenseNova-Skills on GitHub](https://github.com/OpenSenseNova/SenseNova-Skills)
- [README](https://github.com/OpenSenseNova/SenseNova-Skills/blob/main/README.md)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/opensensenova-sensenova-skills