Datapunkt: Skills
Welcome to the Datapunkt Skills guide. This page describes how you can dynamically extend your AI agents' capabilities, customize their intelligence layers, and discover available skills in our repository.
What Are Skills?
In Datapunkt, Skills are modular, dynamic capability packages that provide your AI agents with specialized reasoning, specific domain knowledge, and execution workflows. While our agents (such as the Source Catalog Agent or Data Lineage Agent) possess powerful out-of-the-box analytical capabilities, Skills allow you to inject deep, target-specific expertise directly into their runtime pipeline.
Dynamic Skill Addition
Datapunkt is built with extensibility at its core. You can dynamically add new skills to your agents without redeploying the core platform or modifying agent source code.
This dynamic capability allows you to:
- Extend Capabilities on the Fly: Instantly provision agents with new custom scripts, prompt templates, or integration schemas.
- Tailor Agent Focus: Configure agents to handle specific databases, API interfaces, metadata formats, or proprietary corporate data architectures.
- Collaborative Development: Enable different teams to write and deploy skills independently, which the agent can immediately load and apply in real time.
Describing Skills
Every skill in the Datapunkt ecosystem is self-describing, structured, and version-controlled. When defining a new skill, it is structured to include:
- Metadata & Frontmatter: A unique identifier, name, and concise description of the skill’s purpose.
- Behavioral Rules: System instructions, prompt injection guidelines, and constraints defining when and how the agent should activate the skill.
- Execution Logic: Script files, execution flows, or schemas that provide the agent with concrete tools to interact with your data environment.
This structured description ensures the core orchestrator can dynamically understand the skill's capabilities, determine the optimal scenarios for its application, and explain its actions to the user.
Explore Our Skills Repository
All standard, community, and reference skills for the Datapunkt ecosystem are open-source and hosted in our main git repository. Here, you can find example implementations, structural templates, and core skills that you can customize or deploy.
Go to our Git repository to explore, contribute, or copy existing skills:
👉 Datapunkt Intelligence Repository
Feel free to browse the repository to see how skills are structured, download reference files, or submit pull requests for new skills to share with the community!