Cody autocompletes single lines, or whole functions, in any programming language, configuration file, or documentation.
Every day, Cody help developers write > 25,000 lines of code
High-quality autocomplete must balance speed and accuracy. We’re actively experimenting with new LLMs and context retrieval methods to build the best autocomplete experience.
Embeddings for greater codebase context
Supports all code hosts and multiple LLM options
Zero retention on inputs and outputs, Indemnity
Cody chat helps unblock you when you’re jumping into new projects, trying to understand legacy code, or taking on tricky problems.
Cody can answer questions like:
How is this repository structured?
What does this file do?
Where is X component defined?
Why isn’t this code working??
Write, describe, fix, and smell code with commands.
We’re adding new commands frequently, plus you can create & share your own custom commands.
Cody uses context to answer questions that require an understanding of multiple files or even entire repositories. Plus, this context allows Cody to make suggestions that use your own APIs and idioms.
We’re experimenting with several methods of context retrieval to improve Cody’s accuracy, including embeddings, keyword search, and hybrid search.
Cody context architecture
Context awareness is key to the quality and precision of Cody. This paper outlines how Cody fetches the right context at the right time to answer queries.Read more
Free forever for individual devs on public and private code, with a generous rate limit.