
BAML is a programming language for developers building AI agents, with typed model calls and local tracing built into the language. It runs standalone on macOS, Linux and Windows, or alongside an existing application. The language is open source under Apache 2.0, and works offline.
Its syntax resembles TypeScript, but types persist at runtime and the language excludes any and unchecked casts. Typed errors and static analysis help catch mistakes before a program runs. For model calls, developers define the expected inputs and outputs; BAML parses the response against those types and can repair malformed output. AI functions support streaming, batching, websockets and voice across model providers.
Every function gets a local trace and profile that agents can inspect. This gives developers a record of execution beyond the model call itself. Boundary Web Services is a separate cloud offering; the language, runtime and local tracing run on your own machine.
BAML also includes concurrency controls, retries, cancellation and sandboxing for agent workflows. Its testing framework accepts examples, datasets and production traces, and can grade AI functions across repeated runs rather than rely on a single result.
You can adopt it incrementally. Language bridges let existing projects call BAML functions and share types, generics and lambdas. Supported targets include Python, TypeScript for Node and the web, Go, Rust, Java, C#, C++ and Swift, with Kotlin support for Android and Swift support for iOS.
Claim this page with an email at boundaryml.com. BAML gets the verified badge, and you can upgrade the listing to be featured on localhosted. Proud to be listed? Put our badge on your site.
Want more people to find BAML?Promote it
Something wrong or outdated on this page?
42.4KUpdated 1 day agoApache-2.0
Docker · Web#Guardrails#Human approval#LLM tracing
Agno is a Python framework and runtime for developers building customer-facing or internal AI agents. You can run its agent platform locally with Docker, on your own servers or in your cloud. The open-source framework uses the Apache 2.0 license, and the platform keeps sessions, memory, knowledge and traces in your database.
28.4KUpdated 21 hours ago
Web#Human approval#LLM tracing#MCP
21.3KUpdated 9 months agoApache-2.0
Web#Guardrails#LLM tracing#MCP
11.2KUpdated 5 months agoMIT
VS Code#Code execution#LLM tracing#Visual workflows
29.8KUpdated 23 hours agoMIT
macOS · Windows · Linux#Code execution#Guardrails#Human approval
2.8KUpdated 1 day agoApache-2.0
Windows · Linux · Docker · Web#LLM tracing#Ollama integration#Prompt versioning
Mastra is a TypeScript framework for developers building AI agents and applications on their own servers or inside existing web apps. Its server runs locally or as a standalone deployment; Mastra Cloud provides a hosted alternative. Model routing connects to providers such as OpenAI, Anthropic, and Gemini, so running the framework locally doesn't keep those model requests on your machine.
Rasa is an AI agent platform for product teams building customer-facing text and voice assistants. Teams can deploy agents on their own infrastructure and choose their models and data arrangements. Its CALM engine combines language model understanding with business flows whose code enforces rules, so an assistant can handle conversational wording while following defined processes.
Prompt flow is an MIT-licensed, open-source toolkit for developers who build LLM applications and need to test their behavior before deployment. Its development tools run locally, while an optional cloud version in Azure AI supports team collaboration. Feature development has ended.
OpenAI Agents SDK is an open-source Python framework for developers building AI apps that need to use tools, delegate tasks, or work across multiple steps. Its runtime manages agent turns and conversation state while letting developers express workflows in ordinary Python. It uses the MIT license.
OpenLIT is a self-hosted platform for developers who need to understand how their LLM applications and AI agents behave. It connects model calls with tool activity, retrieval and agent steps, so teams can investigate errors and compare cost, latency and output quality across a workflow.