Urgent.News

What's breaking now, across thousands of outlets.

AI

Introducing bolt.gives: Open-Source Agentic Coding With Any LLM

AI has changed the way we write software remarkably quickly. But I believe there is an important difference between an AI that can write some code and an AI system that can actually build something. That distinction is what led us to bolt.gives. bolt.gives is an open-source Agentic Coding platform that lets you describe an application in plain English and turns that request into a real, working…

Introducing bolt.gives: Open-Source Agentic Coding With Any LLM

AI has transformed software development, but there's a key difference between AI that writes code and AI that builds applications. This led to the creation of bolt.gives, an open-source Agentic Coding platform. bolt.gives transforms plain English prompts into real, functional web applications. It's released under the MIT license.

Instead of just giving you code, bolt.gives actively builds your application. You provide a description like "Build a responsive appointment booking application for a mobile dog groomer," and the system creates the necessary files, installs dependencies, runs commands, starts the development server, and verifies the application's functionality before confirming completion. You can then preview the application and make adjustments, such as modifying the booking screen for a modern look and adding dark mode.

One of bolt.gives' key principles is that developers can choose their preferred AI provider. The platform supports configurable model providers, allowing seamless integration with your favorite AI service. This separation from any specific AI ecosystem ensures the development environment remains versatile as AI models advance.

Beyond just generating code, bolt.gives performs various tasks like project file manipulation, dependency installation, command execution, application running, Preview monitoring, failure recovery, context retention, Supabase integration, and publishing to a Cloudflare-backed address. The platform ensures that the generated code is genuinely functional, not just syntactically correct.

Reliability is a top priority for bolt.gives. While AI can produce plausible-looking code, it often leads to issues like missing dependencies, malformed manifests, or broken imports. To address this, bolt.gives verifies that the Preview can actually load the application, not just the existence of generated files. The system also includes automatic recovery for common failures, attempting to mend the project instead of immediately presenting the problem to the user.

bolt.gives is designed for self-hosting, making it accessible to developers who want to own their development environment. The project can run on Ubuntu without requiring PostgreSQL or SMTP servers, offering flexibility for private installations. The codebase is divided into six major modules: Core, Agent, Runtime, Project, Control Plane, and Surfaces, allowing developers to work on specific areas without needing to understand the entire system.

As an open-source project on GitHub with the MIT license, bolt.gives invites contributions from developers worldwide. The platform welcomes improvements, new features, and innovative ideas. You can explore the source code and give feedback at GitHub: https://github.com/embire2/bolt.gives and try out bolt.gives at https://bolt.gives. Whether you're interested in AI-assisted development, agent architecture, developer tooling, or open-source projects, bolt.gives welcomes your input and contributions.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

More from Thursday 24 September →