SeaGOAT is a local-first semantic code search engine designed for developers to semantically search their codebase using vector embeddings. It addresses the need for efficient, privacy-focused code exploration without relying on external APIs, offering a powerful alternative to traditional keyword-based search.

How It Works

SeaGOAT leverages vector embeddings generated by a local embedding engine and stored in ChromaDB to enable semantic search. It complements this with ripgrep for traditional keyword and regex-based searches, providing a hybrid approach. A local server architecture ensures all processing and data remain on the user's machine, prioritizing privacy and offline usability.

Quick Start & Requirements

• Install via pipx install seagoat.
• Requires Python 3.11+, ripgrep, and optionally bat for colored output.
• Official documentation: https://github.com/kantord/SeaGOAT

Highlighted Details

• Local-first execution with no external API dependencies.
• Hybrid search combining semantic (vector embeddings) and keyword/regex matching.
• Supports common programming languages including Python, JavaScript, TypeScript, C/C++, Go, Java, PHP, Ruby, HTML, Markdown, and text files.
• Integrates with .gitignore for file exclusion and offers custom ignore patterns.

Maintenance & Community

• Project is actively developed by a single maintainer.
• Community support and contributions are welcomed, particularly for macOS and Windows testing.

Licensing & Compatibility

• Licensed under an open-source license (specific license not explicitly stated in README, but implies permissive use).
• Designed for local execution, making it compatible with closed-source projects without data leakage concerns.

Limitations & Caveats

The project is partly tested on macOS and requires community help for Windows support. While designed for local use, self-hosting on a different machine or in the cloud is possible but requires manual security configuration for private code.