$ show OpenSenseNova/SenseNova-Skills
SenseNova-Skills is a composable collection of AI agent skills for office automation, built around the SenseNova model family. It provides production-grade capabilities for image generation and recognition, presentation (PPT) creation, Excel data analysis, deep research with structured evidence, and multi-source search (academic, social, code, finance). The skills are designed to run in Agent Skills-compatible runtimes (OpenClaw, hermes-agent) and are bundled into the Raccoon product for enterprise use. The architecture is modular: each skill is a self-contained directory with a SKILL.md manifest, scripts, prompts, and schemas. The deep research skill implements a multi-agent pipeline with formal JSON schemas and validators. The PPT skill has a multi-stage pipeline from user input to HTML-to-PPTX conversion via Playwright. The codebase demonstrates solid engineering with reusable components like a shared LLM/VLM client, export pipeline libraries, and validation infrastructure. However, it suffers from duplicated utility code across skills, complex environment variable fallback chains, hardcoded paths, and dependency on external APIs and headless browsers. The repo is actively maintained with biweekly reports and frequent commits.
The LLM identified 4 files worth highlighting for reuse:
A self-contained, reusable LLM/VLM/T2I client that wraps OpenAI-compatible chat completions and image generation APIs. It automatically loads environment variables with a fallback chain, supports streaming, and provides a clean interface for agent skills. This module is used by all PPT skills and could be extracted into a standalone library with minimal changes.
A modular Node.js library that converts a DOM intermediate representation (extracted from HTML via Playwright) into pptxgenjs slide objects. It handles complex layouts, text styling, tables, charts, images, and pseudo-elements. The builder is the core of the PPT export pipeline and demonstrates non-obvious skill in mapping CSS properties to PowerPoint OOXML constraints. It is reusable for any project needing HTML-to-PPTX conversion.
An abstract base class for image generation backends (SenseNova, OpenAI, NanoBanana). It defines a clean interface with methods like generate, with_image, and supports configuration via environment variables. The adapter pattern allows easy addition of new backends. This component is isolated and could be reused in any project requiring pluggable image generation.
A validation script that enforces a strict JSON schema for deep research evidence, including claim kinds, source tiers, writing contexts, and cross-references. It uses a reusable helper pattern (err, validate_keys, validate_string_array) that is shared across all deep research validators. This component demonstrates a robust approach to data quality gates in agent pipelines and could be adapted for any structured data validation task.