April 2025 Summaries

Runme` is a project that integrates with `Dagger`, a container-native shell, to provide a more seamless and efficient CI/CD experience. The team at Stateful, which maintains the Runme project, has built a frontend for Dagger Shell into Runme, allowing users to leverage Dagger's powerful primitives to maintain a first-class developer experience. This integration enables features such as portable pipelines, documentation-native workflows, and improved debugging capabilities. By using `Runme` as a Dagger Shell frontend, developers can break down complex pipeline definitions into concise "nouns and verbs" that are native to Markdown, making it easier to manage configurations and secrets for environments. The project aims to reduce coupling with single CI/CD providers while maintaining high developer experience standards.

Dagger is extending its capabilities to integrate Large Language Models (LLM) into its software delivery workflows. This allows developers to build AI-powered automation, such as agents that can perform tasks like code analysis or deployment gates, directly within their workflows. LLMs are now a native component inside Dagger, enabling users to configure models, chain operations, and connect inputs/outputs within existing workflows. The integration provides real-time observability and enables the creation of autonomous, containerized workflows. Examples demonstrate how to build an AI code review agent using Dagger and LLM, showcasing its capabilities for tasks like code analysis. The LLM primitive is available starting in Dagger v0.18 as an experimental feature.

The benefits of using an AI agent to automate fixes in CI pipelines are discussed. The process is made more efficient by reducing manual debugging and fixing routine errors, saving time and context switching for developers. A custom Dagger module encapsulates the specific actions the agent needs, exposing them as tools for a Large Language Model (LLM) to use effectively. This approach leverages Dagger's power for composition and reuse, allowing the agent to reuse existing testing and linting functions directly. The integration of the AI agent into CI transforms the developer experience, providing validated, ready-to-commit suggestions directly in pull requests.

Dagger is a platform that lets you code your software delivery workflows using standard programming languages like Go, Python, and TypeScript. Your pipelines become just code, testable, versionable, and shareable. Your builds run anywhere, locally, in CI, or in Dagger Cloud, ensuring portability. You compose your pipeline from modular functions, backed by a powerful container runtime. Additionally, you can add Large Language Models into your workflows for even greater automation. To help smooth the transition for Earthly users, Dagger is offering one year of free access to its cloud team at no cost, a hands-on migration workshop, and a dedicated help forum.

Docker Model Runner is a new solution that makes it easy to set up and run popular AI models in local development environments, eliminating the need for unfamiliar tools. It integrates with Dagger, allowing developers to create their own local AI agents using a few environment variables. The tool uses Docker containers, but runs models natively on the host machine for better performance. Currently, it only supports Apple Silicon machines, but may be extended to other hardware in the future.