app.build (agent)

⚠️ warning - the repo is not actively maintained anymore. Existing code is provided for research purposes - feel free to fork and use in your experiments.

app.build is an open-source AI agent for generating production-ready applications with testing, linting and deployment setup from a single prompt. This agent relies heavily on scaffolding and extensive validation to ensure high-quality outputs.

There were two generations of this agent:

v1 - Python implementation

Original standalone agent located in ./agent/ directory. It was designed to generate CRUD applications on three stacks: (TypeScript + tRPC + Drizzle + React, Python + NiceGUI + SQLModel, PHP + Laravel). The managed service has been discontinued.

See agent/README.md for setup and some usage instructions.

Work report is available on arXiv (accepted at SANER 2026)

v2 - Rust implementation 🦀

It is located in ./edda/ directory. The purpose of this version was to build a more robust architecture with a focus on data applications (dashboards, analytics, data-driven tools).

Unlike the Python version, it is available not only as a standalone agent but also as a MCP powering your favorite agents (like Claude Code) or being wrapped into custom agents programmatically (see klaudbiusz/cli/codegen.py for the example of using with Claude Agent SDK).

Some tools for bulk app generation and evaluation of this version are available under klaudbiusz (wrappers over Claude SDK and relevant orchestration/glue code).

MCP Installation

Prerequisites:

• OCI-compatible container runtime (Docker, OrbStack, Podman...) must be installed and running for Dagger-based sandboxed execution

curl -LsSf https://raw.githubusercontent.com/appdotbuild/agent/refs/heads/main/edda/install.sh | sh

and attach to your favorite MCP client, e.g. Claude Code:

claude mcp add --transport stdio edda -- ~/.local/bin/edda_mcp

Set Databricks environment variables (required for app generation):

export DATABRICKS_HOST=https://your-workspace.cloud.databricks.com
export DATABRICKS_TOKEN=dapi...
export DATABRICKS_WAREHOUSE_ID=your-warehouse-id

Then start using it with Claude Code:

# Create project dir
mkdir ~/dbrx-sales-by-region && cd ~/dbrx-sales-by-region

# Check the 'edda' MCP is available fo claude
claude mcp list

# Run sample app generation
claude "Create a Databricks app that shows daily sales by region"

To deploy to Databricks (optional, requires write to /usr/local/bin), install Databricks CLI:

curl -fsSL https://raw.githubusercontent.com/databricks/setup-cli/main/install.sh | sudo sh

Citation

If you use this work in your research, please cite our paper:

@misc{kniazev2025appbuildproductionframeworkscaling,
      title={app.build: A Production Framework for Scaling Agentic Prompt-to-App Generation with Environment Scaffolding},
      author={Evgenii Kniazev and Arseny Kravchenko and Igor Rekun and James Broadhead and Nikita Shamgunov and Pranav Sah and Pratik Nichite and Ivan Yamshchikov},
      year={2025},
      eprint={2509.03310},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2509.03310}
}

---

Supported by Neon and Databricks.

Built to showcase agent-native infrastructure patterns. Fork it, remix it, use it as a reference for your own projects.