Ortus has released BoxLang AI Explorer, a local, browser-based catalog of runnable BoxLang AI code samples covering chat, streaming, async requests, tool calling, memory, agents, pipelines, RAG, orchestration, and MCP servers. Each sample includes metadata, source code, guidance, and expected output. A hosted version at ai.boxlang.io/explorer lets you browse and read code without executing requests, while running locally via the BoxLang Version Manager lets you actually call live AI providers such as OpenAI, Anthropic, Gemini, DeepSeek, Grok, Groq, Perplexity, OpenRouter, Mistral, Hugging Face, Voyage, Cohere, and AWS. The project is open source and designed to grow as a living reference, with new samples added as .bxs files.
Table of contents
What It IsWhat's CoveredTry It Live, Run It LocalQuick StartBring Your Own ProviderRunning a Sample DirectlyExplore, Fork, ContributeQuestions this post answers
How do I run BoxLang AI Explorer locally to test AI code samples against a live provider?
Install the BoxLang Version Manager, pin the repo's BoxLang version with bvm install 1.16.0 and bvm use, then install the bx-ai module locally with install-bx-module bx-ai --local and start the server via boxlang-miniserver. Copy .env.example to .env, add your provider API key, and load it into your shell before starting the server; the default config points at OpenAI but supports Anthropic, Gemini, DeepSeek, Grok, Groq, Perplexity, OpenRouter, Mistral, Hugging Face, Voyage, Cohere, and AWS. daily.dev surfaces hands-on guides like this for developers wiring AI providers into their stack.
What AI patterns does BoxLang support example code for?
BoxLang AI samples cover chat and structured responses, streaming, async requests, tools and function calling, memory, agents, pipelines, retrieval-augmented generation (RAG), orchestration, and MCP servers. Each sample is a runnable .bxs file with a metadata block, full source, guidance text, and sample output, organized in a searchable catalog by category and difficulty. Developers evaluating AI frameworks compare pattern coverage like this on daily.dev.