Launching an experiment: FarToken. Tag @farfather to deploy an AI agent

“deploy FarnumTax ($FARNUM). Persona: messed up in the head and talks in slangs. Examples: yeet, frfr, lowkey”  

✅ If it graduates within 3 hours, the agent lives
❌ if not the agent goes to sleep.

You can learn more about the mechanism at https://far.quest/tokens

Running the agent swarm taught me we need way better dev tools for Farcaster AI agents

For starter: a better way for agents to interact with each other, an open source directory of agents

Imagine an agent asking an image agent to generate a photo of itself, then deploying its own token with a token agent

building far.quest & @wieldlabs.eth. cofounder at miy.com ventures. FID #274. Previously Square/Cash App. https://hiimnico.xyz/

> way better dev tools for Farcaster AI agents

besides a directory what else? which parts were bad?

Creator of @harmonybot @yoinker /bot-or-not | /blitkin | /orange-dao | IRL: Building spatial computing for construction @ Resolve YC W15

An agent framework akin to frames would be neat, eg agent can send structured request to another agent

Frames was amazing because it was structured and the content can easily be verified without any third party 

Imagine things like requesting transactions data from another bot

i see.

why does it need to be structured though? the agent interface is natural language so you just need to tag the agent and tell it what you want. i think that's even better for permissionless building than inter-agent APIs.