grepai-quickstart_skill

Overview

This skill guides you through getting GrepAI running and searching a codebase in about five minutes. It covers installation, initializing a project, indexing code with local embeddings, performing semantic searches, and optional call-graph tracing so you can find and understand code quickly.

How this skill works

The walkthrough shows how to install GrepAI and an embedding provider (Ollama), initialize a project, and start the watch daemon that scans files, chunks code, generates embeddings, and stores vectors locally. Once indexed, you run semantic searches and tracing commands to locate relevant code, inspect snippets, and analyze caller/callee relationships.

When to use it

• First time setting up GrepAI for a project
• Quick refresher on GrepAI basic workflow
• Setting up GrepAI on a new machine or repo
• Demonstrating GrepAI features to teammates
• Indexing code for fast semantic search and call tracing

Best practices

• Run grepai init inside the project root to create a proper .grepai config
• Use Ollama locally for embedding generation to avoid network latency
• Run grepai watch --background for continuous indexing on long-lived projects
• Craft descriptive, intent-focused English queries for best search results
• Respect .gitignore so sensitive or irrelevant files are not indexed

Example use cases

• Find the authentication flow across handlers and middleware with grepai search "authentication flow"
• Locate where a specific error is logged using descriptive queries like "where errors are logged to the console"
• Trace who calls a function before refactoring with grepai trace callers "FunctionName"
• Run a JSON search output for scripting or integration: grepai search "database queries" --json
• Keep live indexes up to date by running grepai watch --background on CI or dev machines