Golem 1.5 features - Part 9: Agent skills

Posted on April 18, 2026

Introduction

I am writing a series of short posts showcasing the new features of Golem 1.5, to be released at the end of April, 2026. The episodes of this series will be short and assume the reader knows what Golem is. Check my other Golem-related posts for more information!

Parts released so far:

Skills

Using coding agents has become a standard way of writing code. With Golem 1.5, our application templates include an AGENTS.md and a huge catalog of agent skills describing every detail of creating, modifying and testing agents on the Golem platform.

A bootstrap skill explaining the golem new command is going to be available through skills.sh; then depending on the chosen programming language, a large set of Golem specific skills get unpacked to the application's directory.

Common and per-language skills

The skill catalog is still being finalized for the release, but we will have 10-15 language-independent skills, and about 25-30 additional skills for each language (TypeScript, Rust, Scala and MoonBit).

We are running a benchmark with popular coding agents to see how well they can use these skills and this benchmark will run daily on CI from now on, making sure we can keep up with the rapidly changing ecosystem of coding agents.

Areas covered

The skills cover every aspect of writing and running applications on the Golem platform:

Creating new projects or extending existing ones
Adding dependencies
Configuring applications
Writing new agents
Exposing agents through HTTP and MCP
Communication between multiple agents
Working with databases
Calling external services
Integration with AI providers
Creating webhooks
Using advanced features like transactions, snapshotting, etc
Troubleshooting

Every new feature Golem gets starting from 1.5 will immediately have full support for agentic development.