Science Skills is a collection of agent skills for scientific research tasks spanning genomics, structural biology, cheminformatics, and literature search.
The tool addresses the challenge of enabling AI agents to perform specialized scientific workflows with better grounding and higher token efficiency. It does this by providing structured skill modules that integrate insights from over thirty scientific databases and tools including AlphaGenome, the AlphaFold Database, and UniProt. Each skill consists of a main instruction file with YAML frontmatter, helper scripts, and optional reference documentation that agents can invoke to complete domain-specific scientific tasks.
The tool is designed for researchers and developers building agentic scientific workflows who want to leverage specialized scientific knowledge without building integrations from scratch. It suits projects that need to query multiple scientific databases, perform genomic analysis, work with protein structures, or search scientific literature. The skills are installable via npm and integrate directly with Google Antigravity, where they can be enabled through the application's plugin system. Some skills require API keys for full functionality, though others work without them at reduced rate limits. Users who want to customize existing skills or create new ones should maintain their modifications outside the plugin installation directory to avoid being overwritten during updates.
The project maintains structured skill modules with clear documentation and reference materials for each domain. The tool provides guidance to users for obtaining and configuring required API keys during agent execution. The codebase uses the uv package manager for dependency handling, with automatic installation triggered on first use of a skill.
