SenseNova-Skills

# SenseNova-Skills **English | [简体中文](README_CN.md)**

The SenseNova model family plugs directly into agent runtimes such as [OpenClaw](https://openclaw.ai/) and [hermes-agent](https://github.com/NousResearch/hermes-agent), with the skills in this repository extending the models with concrete, end-to-end office capabilities. In this repository each skill lives in its own directory and declares triggers, capabilities, and execution flow through a `SKILL.md` file, following the [Agent Skills](https://agentskills.io/) convention. The skills cover **image generation & visualization**, **slide-deck (PPT) generation**, **Excel data analysis**, **deep research**, **HTML experiences**, **team collaboration**, and **proactive project tracking** — usable standalone or composed into end-to-end workflows. > **Want to see what it can do?** Check out our [**sn-infographic Gallery**](docs/sn-infographic-examples.md) to explore nearly 100 stunning generation cases and steal their **prompt designs** ! ## Available out-of-the-box in Raccoon The latest SenseNova models and the full Cowork-Skill suite in this repo are bundled into [**Raccoon**](https://office.xiaohuanxiong.com/home), with enterprise-grade security and a zero-setup experience — if you'd rather not provision env, API keys, and runtimes yourself, you can use these capabilities directly through Raccoon. Free trial available — no payment required to get started. Raccoon now ships a full upgrade across product capability and client experience: - **Three core office capabilities, strengthened**: powered by SenseNova 6.7 Flash + Cowork-Skill, data analysis, PPT generation, and task planning each take a step up — covering the full loop from multi-file cleaning/analysis to formal report decks, industry/competitive research, and investment memos. - **New: infographic generation**: built on the SenseNova U1 model, compresses complex data, long reports, and business insights into dense, structured, visual infographics that are easier to digest and share. - **New client + local Agent OS**: the cloud model handles heavy reasoning and multimodal understanding; the local Agent OS sits next to your files, work context, and personal habits — delivering a more personalized, local, and secure AI-native office experience. - **Proven at scale**: chosen by 15M+ individual users and thousands of enterprise customers. > Try it: [xiaohuanxiong.com](https://office.xiaohuanxiong.com/home) ## How to Use These skills are designed to run inside an [Agent Skills](https://agentskills.io/)-compatible agent. - **Recommended runtime**: pair them with **[OpenClaw](https://openclaw.ai/)** or **[hermes-agent](https://github.com/NousResearch/hermes-agent)**. - **Recommended LLM**: pair them with the **SenseNova Platform API**. International: [platform.sensenova.ai/docs](https://platform.sensenova.ai/docs), Base URL `https://token.sensenova.ai/v1` Mainland China: [platform.sensenova.cn/token-plan](https://platform.sensenova.cn/token-plan), Base URL `https://token.sensenova.cn/v1` - **Install & configure**: follow the full walkthrough in **[`INSTALL.md`](INSTALL.md)**. > This repo documents both the international and mainland China SenseNova API flows. Make sure the docs page, API key, base URL, and model name all come from the same region. **Recommended: let the agent install the skills for you.** Hand it the repo URL and ask it to clone and drop the skills into the right directory — for example: > *"Please install SenseNova-Skills from https://github.com/OpenSenseNova/SenseNova-Skills into your skills directory."* After it finishes, **you may need to manually restart the agent service** before the new skills are picked up. | Agent | Target directory | |-------|------------------| | [OpenClaw](https://openclaw.ai/) | `~/.openclaw/skills/` | | [hermes-agent](https://github.com/NousResearch/hermes-agent) | `~/.hermes/skills/` | Prefer to install manually? Clone this repository, then copy the subdirectories under `skills/` into the target directory yourself: ```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, swap the target to `~/.hermes/skills/`. Per-category Python dependencies, API keys, and invocation examples are documented in the Full guide for each section. ## Skills List ### Image & Visualization Full guide: [`docs/sn-image-generate_en.md`](docs/sn-image-generate_en.md) (prerequisites, Quick Start, API config, and invocation samples). | Name | Label | Description | | -------------------------------------------------- | ------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- | | [`sn-image-doctor`](skills/sn-image-doctor/SKILL.md) | Environment Doctor | Validates the SenseNova-Skills environment — checks `sn-image-base` install, Python deps, and required env vars; interactively fills missing values into `.env`. | | [`sn-image-base`](skills/sn-image-base/SKILL.md) | Image Base Layer (Tier 0) | Low-level tools — text-to-image (`sn-image-generate`), image editing (`sn-image-edit`), image recognition (`sn-image-recognize`), and text optimization (`sn-text-optimize`) — exposed through a unified `sn_agent_runner.py`, designed to be called by upper-layer skills. | | [`sn-infographic`](skills/sn-infographic/SKILL.md) | Infographic Generation (Tier 1) | Auto prompt-quality scoring, layout/style selection (87 layouts / 66 styles), multi-round generation with VLM review and quality ranking, producing publication-ready infographics. Supports SenseNova U1.5 Lite, including native 4K output. | | [`sn-image-imitate`](skills/sn-image-imitate/SKILL.md) | Image Imitation (Tier 1) | Given one reference image and a target content prompt, generates a new image that imitates the reference. | | [`sn-image-resume`](skills/sn-image-resume/SKILL.md) | Resume Image Generation (Tier 1) | Given resume information, generates a resume image. | ### Presentations (PPT) Full guide: [`docs/sn-ppt-generate.md`](docs/sn-ppt-generate.md) (prerequisites, Quick Start, API config, and invocation samples). | Name | Label | Description | | ---------------------------------------------- | ---------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | [`sn-ppt-entry`](skills/sn-ppt-entry/SKILL.md) | **PPT Entry Point** | **Unified entry point for PPT generation.** Asks the user to choose fast, standard, or creative mode, then collects role / audience / scenario / page count. For standard mode, also asks about image sourcing (AI, web search, or none) and chart rendering (U1 infographics or ECharts). Parses uploaded pdf / docx / md / txt, emits `task_pack.json` + `info_pack.json`, and dispatches to the chosen mode. | | [`sn-ppt-doctor`](skills/sn-ppt-doctor/SKILL.md) | PPT Environment Doctor | Environment check for the PPT pipeline — validates `sn-image-base`, API keys, the Node runtime, and optional deps; writes missing required vars into `.env`. | | [`sn-ppt-creative`](skills/sn-ppt-creative/SKILL.md) | PPT Creative Mode | One full-page 16:9 PNG per slide, generated via `sn-image-generate` with a per-page composed prompt. Falls back to web image search when T2I generation fails. | | [`sn-ppt-standard`](skills/sn-ppt-standard/SKILL.md) | PPT Standard & Fast | `style_spec` → outline → asset plan + per-slot images + VLM QC → per-page HTML → per-page review → PPTX export. Fast mode builds a complete draft immediately with autonomous decisions, then provides structured refinement suggestions. Supports AI-generated infographics (U1) for diagrams and web image search (Serper) for real photos. | ### Data Analysis (DA) Full guide: [`docs/sn-data-analysis.md`](docs/sn-data-analysis.md) (prerequisites, Quick Start, API config, and invocation samples). | Name | Label | Description | | ------------------------------------------------------------------ | ------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | [`sn-da-excel-workflow`](skills/sn-da-excel-workflow/SKILL.md) | Excel Analysis Orchestration | End-to-end Excel pipeline — multi-sheet read, large-file detection (≥10k rows triggers Parquet), cleaning, conditional filtering, cross-sheet aggregation, and Excel/CSV export. | | [`sn-da-image-caption`](skills/sn-da-image-caption/SKILL.md) | Image Understanding & Data Extraction | For image-first inputs — table OCR, chart understanding, screenshot/UI description; parses captions into DataFrames, recreates visualizations, exports Excel/CSV. | | [`sn-da-large-file-analysis`](skills/sn-da-large-file-analysis/SKILL.md) | High-Performance Large-File Analysis | Streaming reads for ≥10k-row Excel datasets (openpyxl read_only + iter_rows), Parquet conversion, memory optimization, chunked processing, large-file writes. | ### Deep Research Full guides: [`docs/sn-deep-research.md`](docs/sn-deep-research.md) and [`docs/sn-deepresearch-cli.md`](docs/sn-deepresearch-cli.md) (prerequisites, Quick Start, CLI setup, and per-stage invocation). | Name | Label | Description | | -------------------------------------------------------------------- | ------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- | | [`sn-deep-research`](skills/sn-deep-research/SKILL.md) | **Deep Research Entry Point** | **Mode-aware deep-research orchestrator** with parallel research work packages, one-pass quick/normal synthesis, and an auditable heavy workflow, producing final `report.md`. | | [`sn-deepresearch-cli`](skills/sn-deepresearch-cli/SKILL.md) | Deep Research CLI | Installs and operates the standalone `sensenova-skills-deepresearch` CLI, coordinating search, research, monitoring, recovery, and report export through a selected Harness or Agent. | | [`sn-research-report`](skills/sn-research-rep