Developer Burak Karakan has released DAC, an open-source dashboard-as-code tool specifically designed for AI agents to create, validate, and deploy data visualizations. The project addresses a fundamental problem: most dashboard tools are UI-driven, making it nearly impossible for agents to build and review visualizations programmatically.
Developer Built DAC After Agents Kept Reinventing Dashboard Infrastructure
Karakan shared his motivation in an April 29, 2026 Show HN post: "When agents became a reality one of the first things I wanted to do was to automate building dashboards. The first, and the most obvious, wall that I ran into was that a lot of the tools were just driven by UI. This meant that without the agents handling browser UIs and whatnot, it wasn't possible to have the agents do that."
His initial approach—having agents build React apps for each dashboard—worked for simple cases but quickly revealed problems. Every dashboard turned out different, required implementing backends to centralize query execution, lacked standardized visualization rules, and had no way to integrate a semantic layer. "In the end, agents ended up reinventing the wheel for every new dashboard," Karakan explained.
Agent-First Design Prioritizes Version Control and Human Review
DAC reimagines dashboards with agents as the primary user. The tool's design principles include:
- Everything driven by version-controllable text files (YAML and TSX)
- Changes easy for humans to review and understand
- Code-driven dynamic content leveraging agents' coding strengths
- Static analysis as first-class citizen for pre-deployment validation
- Standardized deployment based on files in folders
- Built-in semantic layer for metric consistency
Two Approaches Support Different Complexity Levels
DAC offers both YAML-based dashboards for declarative configuration and TSX/React dashboards for programmatic control. The TSX approach allows dynamic charts, conditional rendering, loops, and load-time query generation—all while remaining auditable through version control.
The tool runs as a single binary written in Go. Running dac init creates sample dashboards based on DuckDB. DAC supports 10+ SQL backends including Postgres, MySQL, Snowflake, and BigQuery. Technical features include:
- Validation system for pre-deployment checks
- Custom themes for consistent branding
- Semantic layer ensuring metric definitions stay consistent
- Built-in AI agent ("Codex") for live chat-based dashboard updates
- Integration with Bruin CLI for query execution
Pre-Configured Agent Skills Enable Standardized Dashboard Generation
DAC includes pre-configured authoring skills for Claude and Codex, allowing agents to generate production-grade, standardized dashboards without browser automation or custom BI infrastructure. Dashboards are defined in folders and deployed as a single binary.
The project launched on Hacker News with 59 points and 14 comments, indicating strong community interest. The open-source repository is available on GitHub at bruin-data/dac.
Key Takeaways
- DAC is an open-source dashboard-as-code tool built specifically for AI agents to create, validate, and deploy data visualizations programmatically
- Created by developer Burak Karakan after discovering agents reinvented dashboard infrastructure for each new project when using traditional React approaches
- Supports both YAML-based declarative dashboards and TSX/React programmatic dashboards, both version-controllable and human-reviewable
- Runs as single Go binary with support for 10+ SQL backends including Postgres, MySQL, Snowflake, and BigQuery
- Includes built-in semantic layer, validation system, and pre-configured agent skills for Claude and Codex to generate standardized dashboards