Machtiani can find a needle in a haystack of a repo. It relies on a local embedding model, and amplifies the dimensional space of the project data to hard to calculate heights. And it does it cheap. Use local LLM for your prompt, or any provider such OpenRouter that adheres to OpenAI Chat Completion form.
Codex + MCT is an embedding match made in heaven.
Machtiani can find a needle in a haystack of a repo. It relies on a local embedding model, and amplifies the dimensional space of the project data to hard to calculate heights. And it does it cheap. Use local LLM for your prompt, or any provider such OpenRouter that adheres to OpenAI Chat Completion form.