Ship smarter agents in production with dbt Agent Skills

The development and deployment of coding agents have advanced significantly since the introduction of tools like Claude Code, Opus 4.5, and GPT 5.2, pushing software engineering into a new era where these agents can operate across the entire Software Development Life Cycle (SDLC). However, these agents often struggle with data work, specifically when interacting with dbt projects, due to a lack of understanding of data lineage, iterative query execution, and contract management. To address these limitations, dbt Agent Skills have been developed, incorporating analytics engineering best practices into agents, enabling them to handle data-specific tasks effectively. These skills, packaged as markdown and scripts, provide agents with the necessary context to perform tasks such as building models iteratively, writing unit tests, and managing Directed Acyclic Graphs (DAGs). Factory's experience with implementing dbt Agent Skills highlights the potential for small teams to achieve significant outcomes by leveraging these skills alongside custom organization-specific skills. This approach not only enhances the agents' efficiency in data operations but also scales the development process, demonstrating the power of integrating dbt Agent Skills in modern data engineering workflows.