MachineMetricAgent

As blockchains are expanding to new horizons, like off-chain processes, it is becoming increasingly more difficult to assert the correct work is being done. For example, in robotics, how can blockchains deterministically decide if a robot acted correctly in a massively dynamic environment. Moreover, how can you assert the robot has not been altered by a adversarial engineer or entity? Even scarier, the robot altering itself.

At its core, same as we need proof-of-personhood, we need proof of machinehood which is derived by distributed consensus that a machine's software stack is genuine.

In the small (broken - I am not a Rust dev ~.~) code sample, you'll see pieces of how I was intending to build a metric agent that publishes telemetry which has been signed by proof-of-machinehood. This is a very basic case, but would be incredibly useful.

In addition to robotics, this proof-of-machinehood concept and application would be critical for EigenLayer. Currently, AVSs must implicitly trust operators are benevolent actors and not altering their software package or sniffing user data etc. This project aims to make a small step towards this using Automata's attested proof-of-machinehood.

It intends to use an SGX library to interact with a trusted enclave for proof-of-machinehood generation/signing of metrics. It primarily includes a large README which goes into a long discussion of the product/technical direction for how I see this being used as a decentralized layer broadly used in software stacks and how it could potentially erode some of the cloud provider moat.