Testnet only. Not audited. These pages describe intended behaviour, not shipped behaviour.

Maynexdocs

Overview

Constrained execution for autonomous agents on Robinhood Chain.

Maynex is a policy layer for accounts operated by autonomous agents. An agent holds the ability to act; the policy module decides which of those actions the account will carry out.

Read this before anything else

Everything on this site describes intended behaviour. It is not a description of running code. The contracts are not written yet, the project targets a testnet, and nothing here has been audited.

The problem

An agent that can sign transactions can do anything the account can do. Giving it a key and hoping the prompt holds is not a control. What is wanted is a narrower grant: this agent may move up to this much, to these places, within this period, and anything else requires a human.

That grant has to be enforced where the value moves — in the account itself — rather than in the software deciding what to ask for.

What the module is meant to do

ConstraintIntent
Spend limitsCap value moved per period
Destination rulesRestrict which addresses can be reached
TimelocksDelay large or unusual operations so a human can intervene
RevocationLet an operator withdraw the grant immediately

Each of these has an open problem attached to it. They are listed on the security page, and each has an issue in the repository.

Where to go next