On April 10, 2026, we released clip4llm v1.1.0. This update marks a significant step forward in how you prepare your codebase for LLMs, introducing a more flexible and granular approach to configuration.

What’s new

The headline feature of this release is Scoped Configuration. Previously, clip4llm relied on a single configuration file at your project root. Now, you can place .clip4llm files anywhere in your directory tree.

This creates a layered system where configurations are merged as the tool traverses your project. While global settings in your home directory and root project settings still apply, any .clip4llm file found in a subdirectory can now override or augment the rules for that specific folder and its children. For example, you can maintain a broad exclusion list at the root but explicitly include specific documentation files in a nested docs/ folder.

Why it matters

When working with large-scale projects, a one-size-fits-all configuration often leads to a trade-off: either you include too much and waste precious tokens, or you exclude too much and leave the LLM lacking critical context.

Scoped Configuration solves this by giving you precision. You can now tailor the context for different modules of your application independently. By moving configuration closer to the code it describes, you ensure that the LLM receives exactly what it needs—and nothing more—reducing noise and improving the quality of the AI’s responses.

Getting Started

Upgrading to v1.1.0 is seamless. This is a non-breaking release, meaning your existing root configurations will continue to work exactly as before. To start using scoped rules, simply download the latest binary and begin adding .clip4llm files to your subdirectories.

This post was AI-generated using the model unsloth/gemma-4-31B-it-GGUF:UD-Q5_K_XL. Based on the clip4llm repository and release v1.1.0 on 2026-04-10. Author: release-storyteller