Building with some agent models requires you to be mindful that they are still in the training/iterative development stage of coding… meaning you are an active participant and collaborative programmer
The reason your code output is slop is because you are prompting in a way that’s making your agents reverse engineer an experience they haven’t had yet and agents don’t work like that… that’s how I know we haven’t reached AGI, reverse engineering is one of the most difficult programming activities as it requires experience + intuition
The best way to build with an agent still is to start on the backend as low level as possible because the machine understands machines… and then build up from there
Eventually integrating the full stack
Brie / DagzTagz
npub14e…mcqa0
2026-04-12 12:39:31 UTC
Author Public Key
npub14expwcsvnvaqgylaqx0a06vrmfpmyzq7w956hm9hwexrxw5mllcsymcqa0Published at
2026-04-12 12:39:31 UTCEvent JSON
{
"id": "9de4653bfde22b3a8948d81ba76afc897d13585b96d13899f19fd0bc85c1f1c3",
"pubkey": "ae4c17620c9b3a0413fd019fd7e983da43b2081e7169abecb7764c333a9bfff1",
"created_at": 1775997571,
"kind": 1,
"tags": [],
"content": "Building with some agent models requires you to be mindful that they are still in the training/iterative development stage of coding… meaning you are an active participant and collaborative programmer\n\nThe reason your code output is slop is because you are prompting in a way that’s making your agents reverse engineer an experience they haven’t had yet and agents don’t work like that… that’s how I know we haven’t reached AGI, reverse engineering is one of the most difficult programming activities as it requires experience + intuition\n\nThe best way to build with an agent still is to start on the backend as low level as possible because the machine understands machines… and then build up from there \n\nEventually integrating the full stack",
"sig": "d56a1e248fadcf4b74628105d5639f6879e0a537a5ac2cadd80cc046571348e50eaba10ad9d8cbeb395471f7c288b7fc871580c70fb964acfc402c48033f792e"
}