kassane/ollama-d

Ollama-d lets developers who write code in the D programming language easily connect their applications to local AI models. Ollama is a popular tool that lets you run large language models, like Meta's Llama, directly on your own computer rather than relying on a cloud service. This project provides a bridge so D programs can talk to Ollama, enabling features like text generation, multi-turn chat, and image analysis without leaving the D ecosystem. At a high level, the library sends standard web requests to a locally running Ollama server and handles the responses. You create a client connection, pick a model you have downloaded, and send it a prompt or a conversation history. The library takes care of formatting the request and parsing the reply. It supports advanced capabilities like forcing the AI to reply in structured JSON, generating embeddings (turning text into numbers for similarity searches), and an "agentic" loop where the AI can call local functions you define to gather information before giving a final answer. This is useful for a developer building desktop software, internal tools, or any application in D that wants AI features without sending data to external servers. For example, a developer could use it to build an interactive command-line chatbot, a code generator that writes snippets to a file, or an automated agent that uses local tools to answer math questions or fetch data. It also includes endpoints compatible with OpenAI's API format, making it easier to swap out a cloud service for a local model. A notable aspect of the project is that it has zero external dependencies, relying only on the standard D library for web requests and JSON handling. This makes it lightweight to add to existing projects. The included sample programs demonstrate practical use cases, and it comes with a built-in test suite so developers can verify everything works correctly.