nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpquz2radjw4syfc86dgh47sdgw7ahqrs50q86cyt27vp500x5cvujs2c7qja (nprofile…7qja) nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpq9v7lz5n67uygmsr24rsw9arejnfxhwa3npkk00qpe8tx5ujdtpgs0ask9g (nprofile…sk9g) FWIW the process I have used for large tasks with LLM programming looks like:
1) Use Cursor. It behaves far less quirky than VS Code for programming in my experience.
2) Explain what I want in detail, and ask it to create a high level plan file.
3) Then ask it to break the plan into more detailed subplans, and address any gaps until it is "confident" in the plans, and to save those plan files in the repo, so the work can be deferred to later. Repeat until it says it is confident.
Mitch Downey :pci:
npub1yf…w6g6v
2026-06-01 03:35:49 UTC
in reply to nevent1q…n737
Author Public Key
npub1yfr5d05qj3qzz84kntze3c8accyz0unwczgm0fxuypuqdnenkwuqew6g6vSeen on
wss://relay.ditto.pubPublished at
2026-06-01 03:35:49 UTCEvent JSON
{
"id": "dce699f1634f8fc2ee26fa5fbed25bbfa48d55b84db86d83ee832e0d5083738c",
"pubkey": "224746be809440211eb69ac598e0fdc60827f26ec091b7a4dc207806cf33b3b8",
"created_at": 1780284949,
"kind": 1,
"tags": [
[
"p",
"e0943eb64eac089c1f4d45ebe8350ef76e01c28f01f5822d5e6068f79a986725",
"wss://relay.ditto.pub"
],
[
"p",
"2b3df1527af7088dc06aa8e0e2f47994d26bbbb1986d67bc01c9d66a724d5851",
"wss://relay.ditto.pub"
],
[
"p",
"aef28d9b7d04ff0ce0553a091854068c388884b8c2c0b65f8fc3e25cd9c5b078",
"wss://relay.ditto.pub"
],
[
"p",
"b7fe4910768fb0b39d42bee0b6b2fa60a51a6906bffd0565a144b0b8a6e460b9",
"wss://relay.ditto.pub"
],
[
"e",
"0be0f5a4a1a230e2b24d5ed8bd2a839c5cfbd0cc998bfdcdef200b748b34b13a",
"wss://relay.ditto.pub",
"reply"
],
[
"proxy",
"https://podcastindex.social/users/mitch/statuses/116672754436660088",
"activitypub"
],
[
"client",
"Mostr",
"31990:6be38f8c63df7dbf84db7ec4a6e6fbbd8d19dca3b980efad18585c46f04b26f9:mostr",
"wss://relay.ditto.pub"
]
],
"content": "nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpquz2radjw4syfc86dgh47sdgw7ahqrs50q86cyt27vp500x5cvujs2c7qja nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpq9v7lz5n67uygmsr24rsw9arejnfxhwa3npkk00qpe8tx5ujdtpgs0ask9g FWIW the process I have used for large tasks with LLM programming looks like:\n\n1) Use Cursor. It behaves far less quirky than VS Code for programming in my experience.\n\n2) Explain what I want in detail, and ask it to create a high level plan file.\n\n3) Then ask it to break the plan into more detailed subplans, and address any gaps until it is \"confident\" in the plans, and to save those plan files in the repo, so the work can be deferred to later. Repeat until it says it is confident.",
"sig": "fea00c861e2b7743cbddae2e5a551914f0be8530272793bc77b0c67fb3f1abf3346785fb6facbc8422db9f80c666102e4ca1df2d132da185092d8cffe70fa8e9"
}