With wide enough support, a couple of benefits would be nefarious and malicious components would get highlighted quickly, and it could serve as a feedback channel from consumers to suppliers on why an agreement was rejected.
Ultimately, the power dynamic needs to be recalibrated.
A core component would be standardising clauses so they could be handled individually and automatically. If you’ve already answered the clause, and the parameters (time spans, quantities) were within the range you’d set, it would be ‘green’ and could be hidden. Clauses which you had answered but are outside your criteria would be ‘amber’, and unanswered clauses would be ‘red’. As you process more agreements, your can save answers to your profile so the process becomes more optimised over time.
One concern I'd have is that the level of access I'm willing to give websites isn't always universal, even within the same category. On the other hand, there's some things I really just don't care about and whatever website can help themselves. Just something to consider.
Another component would be the ability to rigorously diff agreements as they change, only those components that change or are in some way dependant are presented for attention.
But this begs the question: Apple Privacy Labels "caught on" because Apple has unilateral control to enforce them in the App Store. If ostensibly the same idea for the WWW did not catch on, is the problem (1) the lack of enforcement/economic incentive mechanisms on the decentralized web or (2) that consumers really didn't care/know enough to create/enforce such free market incentives?