Skip to content

Latest commit

ย 

History

296 Commits

Folders and files

NameName
Last commit message
Last commit date
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 

Repository files navigation

๐Ÿ‰ Loong - Rust Base for Vertical AI Agents

Loong

"Originated from the East, here to benefit the world"

English | ็ฎ€ไฝ“ไธญๆ–‡

Important

This repository is an independent-history rewrite of Loong. It currently contains an architecture and contract skeleton, not a usable agent product.

Loong is a Rust foundation for vertical AI agents. Its priorities are security, performance, extensibility, and long-term evolution. Agents should work reliably in real environments over the long term and collaborate with developers beyond a single demonstration.

Why Loong

Architecture-governed safety

Sensitive operations should be authorized before they happen. Loong models outbound effects as Actions and authorizes them before execution. Policy authorization is the primary boundary; sandboxes and model review can still add defense in depth.

Performance starts with the foundation

Loong uses Rust. Core execution paths should stay fast and runtime overhead should remain predictable. Concrete performance claims require representative workload benchmarks.

Extend it at your own pace

Loong is intended to support two ways to extend the product:

  • Build deep or performance-sensitive integrations in Rust and compile them with Loong.
  • Load scripts and similar extensions at runtime for faster iteration without rebuilding the host application.

The same security expectations apply to both. The exact runtime extension system is still being designed.

Built to keep evolving

The rewrite deliberately begins with a small foundation. New product areas will be added as their responsibilities and safety boundaries become clear, keeping Loong understandable as it grows.

The Actor runtime and its lifetime model make components easy to run, extend, stop, and restart.

Current Status

This rewrite is at an early foundation stage. It is not ready to install or use as a complete agent product yet. The repository will describe capabilities as they become real rather than presenting planned work as finished.

Developer information The technical design is documented in [ARCHITECTURE.md](ARCHITECTURE.md), and unresolved design decisions are tracked in [docs/open-questions.md](docs/open-questions.md). The [release guide](docs/development/releasing.md) documents the crates.io workflow for `loac` and `loac-macros`.

The current checks are:

cargo fmt --all -- --check
cargo check --workspace --all-targets
cargo test --workspace

License

Loong is licensed under the MIT License.

About

Lightweight, clear, and fully extensible AI agent infrastructure โ€” learn easily, customize anything ๐Ÿ‰

Topics

Resources

Stars

638 stars

Watchers

7 watching

Forks

Releases

Packages

Contributors

Languages