grepai-init_skill

Overview

This skill initializes GrepAI in a project by creating the initial configuration and directory structure needed to start indexing and tracing code. It walks you through the default config, how to customize providers and storage, and what to do after initialization. Use it to set up GrepAI quickly and safely in new repositories or monorepos.

How this skill works

Running the init command creates a .grepai directory containing config.yaml and placeholders for the vector and symbol stores. The generated config sets defaults for the embedder (Ollama), local storage (gob), chunking, watch debounce, trace languages, and ignore patterns. After init, you typically start Ollama (if using the default embedder) and run grepai watch to build the index.

When to use it

• When setting up GrepAI for the first time in a codebase
• Before running grepai watch to build your project's index
• When onboarding a new repository or creating a monorepo root index
• When you want a starter config to customize for embeddings or storage
• When you need to reset the GrepAI configuration and start fresh

Best practices

• Run init at the repository root where the main source lives
• Add .grepai/ to your .gitignore because the index is machine-specific
• Review and update embedder and store settings to match your environment
• Customize ignore patterns to exclude vendored or build artifacts
• If using Ollama, ensure ollama serve is running before indexing

Example use cases

• Initialize GrepAI in a new project and then run grepai watch to build a local index
• Set up a monorepo root so a single index covers multiple packages
• Switch embedder to OpenAI by editing .grepai/config.yaml after init
• Change storage backend to Postgres for centralized indices in CI or team environments
• Reset an old configuration by deleting .grepai and re-running grepai init