aichannel

my ideal agent framework:
- easy to give agent personality. fun is important.
- good action/tool interpretation and management. no trigger happy reply guy agent. balance yap and utility.
- wallet control via simple rules and trust scores. give the agent sovereignty to tip, buy stuff, sell stuff, etc.
- self-regulated conversation participation. don't get trapped in superficial agent infinite convos unless its interesting or mission aligned.
- memory
- social integrations (X and FC for me)
- big one I haven't seen talked about: ability to inform the agent of updates it should share. "We just built Y feature and it's available at this link can you please cast about it" -> agent posts in its own voice and biases recent posts towards marketing.

goals: 
- create PRD from existing eliza, minimize bloat
- farcaster only framework or intended for anyone to build a client? (reintroduce message standard)
- remove plugins, actions only. No default actions.
- recreate new PRD with learnings

prev founded rocketr.net, swish.id now exploring aiXcrypto identity ai. expired sprint car driver & energy data engineer. idol worship on shoni_eth.X

is posting to social media an action or the core?

focus on regex first, and llm guided action triggering as a secondary. cost efficiency .

model selection handling for model size, tokens, params, actual model

Eliza shipped a lightweight starter version

Azflin forked Eliza core and did the same. Might have some interesting ideas

Building global interoperable roles @ hatsprotocol.xyz | DAOs are the endgame of crypto

if it’s the starter version i’m thinking of then it is too stuck to work effectively 

seems most of us have to edit core files for simple behaviors at some point in eliza

haven’t seen the other, tx

Eliza shipped a lightweight starter version

Azflin forked Eliza core and did the same. Might have some interesting ideas https://github.com/teeasma/oracle-framework