About God Mode Search
The prior-art search that GitHub never built.
The problem
GitHub has over 395 million public repositories and a keyword search engine. That combination creates a paradox: the answer to almost any software question exists somewhere in the corpus, but finding it requires already knowing the vocabulary used by the people who built it.
A search for "tool that helps roommates split household chores" will never surface a repository whose README says "collaborative task management for shared living spaces" — even though they describe the same product. This is the vocabulary mismatch problem, and it is structural. No amount of clever keyword generation or synonym injection fully solves it, because the mismatch isn't about synonyms — it's about concepts, intentions, and use cases expressed in fundamentally different language.
Our approach
God Mode Search maintains a continuously updated semantic index over the largest feasible subset of GitHub's public repository corpus. We read every repository's name, description, and README, convert them into a high-dimensional vector that captures their meaning, and store those vectors in a searchable index.
When you describe what you're looking for in plain language, we convert your description into the same kind of vector and find the repositories whose vectors are closest to yours. The result is a ranked list of real GitHub projects that occupy the same semantic neighborhood as your query — even when your words and theirs share nothing in common.
What we are
We are a project discovery tool. Given a natural-language description of a concept, use case, problem, or solution, we return a ranked list of GitHub repositories that occupy the same semantic space. We search across projects by what they do, not how their code is written.
What we are not
Not a code search engine. We don't search inside source code for functions, classes, or patterns. Sourcegraph, GitHub code search, and grep.app serve that use case.
Not a clone detector. Two projects can be semantically identical in our index — they solve the same problem — while sharing zero lines of code.
Not a recommendation engine. We don't analyze your stars, contributions, or history. Results are determined by the query, not the user.
Not a quality judgment. We rank by semantic similarity to your query, not by project quality. You still need to evaluate maintenance status, license compatibility, and fit for your needs.
Coverage & transparency
Our index currently covers over 34 million repositories with more than 820,000 fully embedded READMEs. These numbers grow daily as we continuously ingest new repositories and embed their content. Our goal is to index every public repository with meaningful content, and we publish our coverage stats openly.
A note on licensing
A repository appearing in our results means it is publicly visible on GitHub. It does not mean the code is free to use, modify, or redistribute without conditions. GitHub hosts repositories under every imaginable license — from fully permissive (MIT, Apache 2.0) to highly restrictive (GPL, AGPL, proprietary source-available). Users are responsible for reading and complying with each repository's license before using its code in any context.
The bet
The open-source ecosystem has grown so large that discovery has become the bottleneck. The problem is no longer "does a solution exist?" — it almost certainly does, somewhere among hundreds of millions of repositories. The problem is "can I find it?" And the answer, with keyword search, is often no.
We believe a high-quality semantic index over this corpus, searched by meaning rather than vocabulary, is valuable enough that individuals will use it, organizations will pay for it, and the dataset itself will become a reference asset for the developer ecosystem.