Content
@
https://warpcast.com/~/channel/devfc
0 reply
0 recast
0 reaction
Samuel ツ
@samuellhuber.eth
Observations from 20+ hours with cursor agent searching the web, checking files and using cursor rules. Agents run very well if there is structure. They struggle without it. Guardrails they can operate within make them highly capable. - provide context in README files they know to check and let them keep them up to date - provide cursor rules file (you may need to restart the agent or interrupt via chat to remind them of checking rules) - having an extensive test suite is awesome! the agent will self check! just make sure it doesn't adjust anything to just make tests pass (that's where README's and rules come in, though agents are NOT deterministic! You have to review) - having strong linting makes the agent reevaluate and adjust at runtime before even running tests which is way faster! Strong typing and styling rules in IDE come in clutch here! haven't yet tested with MCP enabled that may even pull in whole business logic as reference from external docs / chats?
2 replies
1 recast
14 reactions
Michael Pfister
@pfista
fastest and easiest way to try MCP is with skeet.build :)
0 reply
0 recast
2 reactions
Juliettemeon 🎩
@juliettemeon.eth
Hi samuell, can cursor agent be used in daos? Is it capable of tracking and monitoring the activities of dao members ?
0 reply
0 recast
0 reaction
