Grepai Init

Initializes GrepAI's semantic code search in your project by creating a .grepai directory with config.yaml and setting up sensible defaults. You get Ollama with nomic-embed-text for local embeddings, 512-token chunks with 50-token overlap, and ignore patterns that skip node_modules, .git, and other usual suspects. The config supports multiple embedding providers (OpenAI, Ollama) and storage backends (local gob files or Postgres). One nice touch is the trace mode configuration that enables symbol indexing across 20+ languages out of the box. Just run it once at your project root, tweak the YAML if needed, then start the watch command to build your index.

skills.sh installs: 724
GitHub stars: 20
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add yoanbernabeu/grepai-skills --skill grepai-init --agent claude-code

Installs into .claude/skills of the current project.

Files

GrepAI Init

This skill covers the grepai init command and project initialization.

When to Use This Skill

Setting up GrepAI in a new project
Understanding what grepai init creates
Customizing initial configuration
Troubleshooting initialization issues

Basic Usage

cd /path/to/your/project
grepai init

What Init Creates

Running grepai init creates the .grepai/ directory with:

.grepai/
├── config.yaml    # Configuration file
├── index.gob      # Vector index (created by watch)
└── symbols.gob    # Symbol index for trace (created by watch)

Default Configuration

The generated config.yaml:

version: 1

embedder:
  provider: ollama
  model: nomic-embed-text
  endpoint: http://localhost:11434

store:
  backend: gob

chunking:
  size: 512
  overlap: 50

watch:
  debounce_ms: 500

trace:
  enabled_languages:
    - .go
    - .js
    - .ts
    - .jsx
    - .tsx
    - .vue
    - .py
    - .php
    - .lua
    - .gd
    - .rb
    - .c
    - .h
    - .cpp
    - .hpp
    - .cc
    - .cxx
    - .hxx
    - .rs
    - .zig
    - .cs
    - .java
    - .fs
    - .fsx
    - .fsi
    - .pas
    - .dpr

SKILL.md on GitHub