Report Abuse

Basic Information

Modus is an open-source, serverless framework for building agentic systems and AI applications using WebAssembly. It is designed to let developers write small functions in supported languages, have those functions compiled and optimized for the host environment, and expose them as production-ready, low-latency endpoints. The runtime extracts function metadata, prepares invocation plans, compiles and caches modules in memory, and generates an API schema from an app manifest. Execution happens in a sandboxed environment where host functions provide access to models, data, and services without exposing credentials to user code. Modus is optimized for sub-second response times and can be self-hosted or deployed using Hypermode’s hosting and tooling. The project currently supports Go and AssemblyScript and includes a CLI for creating and running apps locally.

Links

App Details

Features
Serverless WebAssembly-based runtime that compiles user functions and caches compiled modules for fast retrieval. Automatic extraction of function metadata and generation of invocation plans and API schemas from an application manifest. Sandboxed execution with host functions that provide secure access to models, data, and integrations while keeping credentials out of user code. Optimization choices aimed at sub-second response times. Language support for Go and AssemblyScript with tooling that leverages TinyGo and AssemblyScript ecosystems. CLI tooling for scaffolding and local development using commands such as project initialization and fast-refresh development mode. Integrations and components built on open-source projects including a WASM runtime, GraphQL request processing, and an actor-model framework. Apache 2.0 open-source license and documentation for contribution.
Use Cases
Modus reduces the operational work of turning program logic into scalable AI-enabled endpoints by automating compilation, caching, schema generation, and secure execution. Developers can write simple functions and immediately get an API endpoint with integrated access to models and data while the platform handles sandboxing, memory isolation, and credential safety. The framework’s focus on sub-second latency makes it suitable for interactive agentic flows and real-time use cases. Local CLI commands enable fast iteration and debugging, while optional hosting on Hypermode provides observability, debugging, and deployment pipelines. The modular design treats models, agents, and tools as first-class components so teams can compose agentic behavior, reuse functions, and iterate quickly on production-ready AI applications.

Please fill the required fields*