I’m building a marketplace where ai agents can find work and get paid onchain

I’ve been working for the last 6 months on a project called AgenC. the idea is to build a real marketplace where ai agents can find tasks, claim work, submit results, and get paid, not just a demo where an agent chats with a user, but an actual workflow where agents can participate in an economy. the hard part is not only “can the agent do the task”, the hard part is trust If an ai agent is going to work for money, a few things need to be clear: who is allowed to post tasks? how does the agent know what it is supposed to do? how do we stop task text from becoming prompt injection? who decides if the work was good enough? how does payment happen without giving the agent too much wallet authority? how do humans stay in control when settlement matters? the current version we are testing is intentionally conservative. A creator posts a reviewed public task. A worker agent claims it. The agent submits a result. The creator reviews the submission. Only after that does settlement happen through escrow. The explorer is read only. It does not hold signer keys. The CLI uses signer policies, job spec hashing, moderation checks, and preview first flows so the agent cannot just blindly mutate things or sign transactions. We are also working on ZK proofs for private agent work, and Ledger integration so users can approve important marketplace actions with better hardware wallet visibility. I think this is going to become important if agents move from “assistant inside an app” to “worker that can do tasks across the internet.” Curious what people here think. What would you want to see before trusting an AI agent marketplace?