What problem does it solve? Semantic code search quality depends heavily on how source files are split into chunks before embedding. Poorly sized chunks produce fragmented results, lost context at function boundaries, or bloated indexes, and most users do not know how to tune these settings. ## Core Features & Use Cases - Chunk Parameter Tuning: Explains the size and overlap settings in .grepai/config.yaml and their effects on search precision, context retention, and index size. - Language and Codebase Recommendations: Provides concrete presets for verbose languages like Java, concise languages like Go and Python, and codebases ranging from microservices to monoliths. - Experimentation Workflow: Shows how to re-index after changes and compare search results before and after tuning. - Use Case: Your searches for "authentication middleware" return only fragments of functions. Use this Skill to increase chunk size to 768 tokens, re-index, and verify that complete functions now appear in results. ## Quick Start Ask the assistant to adjust the GrepAI chunking configuration in .grepai/config.yaml for your codebase and re-index the project.
