astrid-runtime/sdk-js

JavaScript and TypeScript SDK for building Astrid capsules.

What it solves

It provides a way for developers to build "capsules" (isolated, portable binaries) using JavaScript and TypeScript instead of Rust. This allows developers to use familiar Node.js-like APIs and TypeScript decorators to create modular components that can be executed by the Astrid runtime kernel without needing to change the host ABI.

How it works

The SDK uses a build pipeline that transforms TypeScript code into a WebAssembly (Wasm) component. The process involves:

  1. Typechecking: tsc compiles TypeScript to JavaScript.
  2. Bundling: esbuild bundles the code and the SDK into a single ESM module.
  3. Componentization: ComponentizeJS converts the bundle into a wasm32-wasip2 component using StarlingMonkey.
  4. Packaging: A Rust-based tool packs the resulting Wasm file and a Capsule.toml configuration into a .capsule archive.

Who it’s for

Developers who prefer the ergonomics of JavaScript and TypeScript over Rust for building Astrid capsules, and those who are building daemon-mode workloads where the larger binary size of JS capsules (compared to Rust) is an acceptable trade-off.

Highlights

  • Familiar API: Uses node:fs/promises, node:child_process, and WHATWG fetch.
  • TypeScript Decorators: Uses decorators like @capsule, @tool, and @install to define capsule behavior and lifecycle hooks.
  • Language Agnostic: Produces the same .capsule archive format and WIT contract as the Rust SDK, making binaries interchangeable from the kernel's perspective.
  • Integrated Build Tooling: Includes a build orchestrator that manages the entire pipeline from source code to the final archive.

Related

  • Project
    astrid-runtime/astrid

    Astrid is an operating-system project for WASM-based capsules that provides an OS-grade security boundary for AI agents, separating decision-making from execution authority.

  • Project
    TanStack/ai

    A type-safe, provider-agnostic TypeScript SDK for building streaming chat, tool-calling agents, and multimodal AI applications across various JS frameworks.

  • Project
    vercel/chat

    A unified TypeScript SDK for building chat bots once and deploying them across multiple platforms like Slack, Discord, and WhatsApp, featuring native AI streaming support.

  • Project
    qdrant/qdrant-js

    A JavaScript/TypeScript SDK for the Qdrant vector search engine, providing REST and gRPC clients to manage and search vector data.

  • Project
    google/adk-js

    A code-first TypeScript toolkit for building, orchestrating, and deploying AI agents with full type safety and support for both Node.js and browser environments.