Projects

Products and open infrastructure for AI agents and models.

Everything here is built by Cottus AI. Public projects link to their code and docs; the ones still private say so.

Cottus

Hands and permissions for AI agents on the Linux desktop. Agents you already use work your apps and browser through it, under your policy, with approvals and a full log.

Rust · MIT
Early development, private for now Read about Cottus

Offprint

A studio for short videos that runs on your computer. Claude writes the script and plans every shot, generative models make only the footage, and the paper cut-outs, type, captions and cut are made locally, in 34 edit styles.

Go
Launching soon, private for now Read about Offprint

Trinetra

Continuous security testing that runs inside your own network. It tests the systems you own on a schedule and at every deployment, points to the exact file and line, and can draft the fix.

Rust · Proprietary
Coming soon, private for now Read about Trinetra

akernl

A self-hosted sandbox API for AI agent code. Run Python, Node.js, shell or Go in Firecracker microVMs with one HTTP call.

Python · MIT Overview Code Docs

Cottus Runtime

An LLM inference engine in C++ and CUDA for Llama architectures, with a paged KV cache, its own attention kernels and Python bindings.

C++, CUDA · Apache-2.0 Overview Code PyPI

argus

Lightweight profiling and tracing for Python ML workloads, exported as Chrome traces for Perfetto.

Python · Apache-2.0 Overview Code

barecott

A unikernel runtime for AI agents.

Apache-2.0
Private for now