Repolode
kantord

kantord/SeaGOAT

A local-first semantic code search engine that combines vector embeddings with ripgrep so you can query a codebase in natural language.

Stars
1,321
Forks
97
Open issues
36
Last commit
8 hours ago

What is it?

SeaGOAT is a command-line code search tool for finding code by meaning rather than exact keywords. It runs a local server that indexes a repository with vector embeddings stored in ChromaDB, and it pairs those semantic matches with regular-expression and keyword results from ripgrep. Queries can be plain sentences such as "where are the numbers rounded" or mix in regular expressions.

The tool runs entirely on your own machine and does not call remote APIs by default. Indexing happens in the background, so you can search while files are still being processed; early results come from ripgrep and the warning shows an estimate of how complete the semantic index is. Repository ignore rules from .gitignore are respected, and additional patterns can be set in a YAML config file.

An optional --generative flag sends results to an LLM to enhance answers. It defaults to a local Ollama server, and OpenAI or MiniMax can be configured instead, though cloud providers receive code snippets and must be enabled explicitly.

Key features

  • 01Natural-language semantic search over local code using vector embeddings
  • 02Combines semantic results with ripgrep regular-expression and keyword matches
  • 03Runs a local server; no remote API calls by default
  • 04Search while indexing is still in progress, with an accuracy estimate
  • 05Respects .gitignore and supports extra ignore patterns in YAML config
  • 06Optional LLM-enhanced results via Ollama, OpenAI or MiniMax

Use cases

  • Finding where a behavior is implemented when you do not know the exact function name
  • Exploring an unfamiliar codebase by asking questions in plain language
  • Searching private code without sending it to third-party services
  • Hosting a shared code search server for a team on a private network

Quick start

  1. 1

    Install ripgrep and optionally bat using your system package manager (listed as dependencies)

  2. 2

    Install SeaGOAT with pipx

    pipx install seagoat
  3. 3

    Start the SeaGOAT server on your repository

    seagoat-server start /path/to/your/repo
  4. 4

    Query your repository in natural language

    gt "Where are the numbers rounded"

Things to know

  • Supports only a fixed list of file types such as Python, JavaScript, Go, and Markdown.
  • macOS and Windows support is only partly tested or needs help.

In collections

Topics

FAQ

What is SeaGOAT?
A local-first semantic code search engine that combines vector embeddings with ripgrep so you can query a codebase in natural language. SeaGOAT is a command-line code search tool for finding code by meaning rather than exact keywords. It runs a local server that indexes a repository with vector embeddings stored in ChromaDB, and it pairs those semantic matches with regular-expression and keyword results from ripgrep. Queries can be plain sentences such as "where are the numbers rounded" or mix in regular expressions.
Is SeaGOAT free to use? What license does it have?
SeaGOAT is open source under the MIT license. Check the license terms before using it in commercial products.
Is SeaGOAT still maintained?
Actively maintained. The last commit to SeaGOAT was on Nov 7, 2026.
What can I use SeaGOAT for?
Finding where a behavior is implemented when you do not know the exact function name Exploring an unfamiliar codebase by asking questions in plain language Searching private code without sending it to third-party services Hosting a shared code search server for a team on a private network
How do I install SeaGOAT?
Install ripgrep and optionally bat using your system package manager (listed as dependencies) Install SeaGOAT with pipx: pipx install seagoat Start the SeaGOAT server on your repository: seagoat-server start /path/to/your/repo Query your repository in natural language: gt "Where are the numbers rounded"
Where can I get SeaGOAT?
The source code and installation instructions are on GitHub: https://github.com/kantord/SeaGOAT

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects