A List of Lists That Is Actually Useful
An organized catalog of the 12 prompt lists I keep, the 3 I use daily, the 4 I revisit quarterly, and the 5 I am embarrassed to admit I have.
Why this post existsI have been collecting prompts for about 3 years. The collecting is the part that started as a notes file and grew into a folder of folders, and the folder of folders is the part that I now have 12 lists in. The 12 lists are the part I want to share, because the 12 lists are the part I have been promising to write about for at least a year, and the promising is the part I want to stop doing. The stopping is the part this post is about.The 12 lists are not the kind of "ultimate list of 100 AI prompts" content that does well on Twitter. The 12 lists are the kind of small, focused, organized-for-me lists that work because they are organized for me, and the organized-for-me is the part I want to capture. The capture is the part I want this post to be, because the capture is the part I have not seen anyone else write about. The write-about is the part I want to be the post.The 12 listsHere are the 12 lists, in the order I would create them if I were starting over.1. The "do this 1 thing" list (daily use)The list is 4 prompts, one for each of the 4 kinds of "I need to do this 1 thing" tasks I have. The 4 prompts are: (a) the "I need to write a 1-paragraph summary of this 10-page document" prompt, (b) the "I need to turn these 5 bullet points into 3 sentences" prompt, (c) the "I need to write a 1-line email to decline this meeting" prompt, and (d) the "I need to write a 1-line email to accept this meeting" prompt. The 4 prompts are the part I use every week, and the every-week is the part that makes the list worth having.2. The "debug this output" list (daily use)The list is 5 prompts for the 5 kinds of "this output is almost right but not quite" tasks I have. The 5 prompts are: (a) the "this output is 80% right, fix the 20%" prompt, (b) the "this output is too long, cut it in half" prompt, (c) the "this output is too short, expand it" prompt, (d) the "this output has the wrong tone, rewrite in [X] tone" prompt, and (e) the "this output is missing the 1 thing I asked for, add it" prompt. The 5 prompts are the part I use every day, and the every-day is the part that makes the list worth having.3. The "ship this thing" list (daily use)The list is 6 prompts for the 6 kinds of "I am about to ship this thing, help me do the last 10%" tasks I have. The 6 prompts are: (a) the "write a 1-paragraph changelog" prompt, (b) the "write a 1-page README" prompt (the one from [the README post), (c) the "write a 3-bullet release notes" prompt, (d) the "write a tweet thread for this post" prompt, (e) the "write a LinkedIn post for this thing" prompt (the one from [the LinkedIn cross-post), and (f) the "write a 1-paragraph email to my list" prompt. The 6 prompts are the part I use every time I ship, and the every-ship is the part that makes the list worth having.4. The "research this topic" list (quarterly use)The list is 4 prompts for the 4 kinds of "I need to learn about this topic in 30 minutes" tasks I have. The 4 prompts are: (a) the "give me the 5 most important things to know about [X]" prompt, (b) the "compare [X] to [Y] in a table" prompt, (c) the "give me the 3 contrarian takes on [X]" prompt, and (d) the "give me a 5-question reading list for [X]" prompt. The 4 prompts are the part I use quarterly, and the quarterly is the part that makes the list worth having.5. The "decide between options" list (quarterly use)The list is 3 prompts for the 3 kinds of "I need to decide between A and B" tasks I have. The 3 prompts are: (a) the "give me a pros/cons table for [A] vs [B]" prompt, (b) the "give me a decision matrix for [A] vs [B] with 5 criteria" prompt, and (c) the "give me a 1-paragraph case for A and a 1-paragraph case for B" prompt. The 3 prompts are the part I use when I have a real decision to make, and the real-decision is the part that makes the list worth having.6. The "audit my tools" list (quarterly use)The list is 2 prompts for the 2 kinds of "I need to triage my AI tool stack" tasks I have. The 2 prompts are: (a) the "audit my tool stack, recommend what to keep, kill, or try" prompt (the one from [the AI audit post), and (b) the "audit my tool stack for duplicate functionality" prompt. The 2 prompts are the part I use every 90 days, and the 90-days is the part that makes the list worth having.7. The "fix this writing" list (monthly use)The list is 3 prompts for the 3 kinds of "I need to fix this draft" tasks I have. The 3 prompts are: (a) the "fix the grammar and spelling, do not change the words" prompt, (b) the "tighten this draft, cut 20% of the words" prompt, and (c) the "rewrite this in [my voice profile]" prompt. The 3 prompts are the part I use monthly, and the monthly is the part that makes the list worth having.8. The "explain this concept" list (monthly use)The list is 3 prompts for the 3 kinds of "I need to understand this concept" tasks I have. The 3 prompts are: (a) the "explain [X] like I am 12" prompt, (b) the "explain [X] like I am a senior engineer" prompt, and (c) the "explain [X] with a concrete example" prompt. The 3 prompts are the part I use monthly, and the monthly is the part that makes the list worth having.9. The "generate this artifact" list (monthly use)The list is 4 prompts for the 4 kinds of "I need to generate a specific artifact" tasks I have. The 4 prompts are: (a) the "generate a CSV with 10 rows of fake data" prompt, (b) the "generate a JSON schema for [X]" prompt, (c) the "generate a markdown table for [X]" prompt, and (d) the "generate a SQL query for [X]" prompt. The 4 prompts are the part I use monthly, and the monthly is the part that makes the list worth having.10. The "summarize this content" list (weekly use)The list is 3 prompts for the 3 kinds of "I need to summarize this" tasks I have. The 3 prompts are: (a) the "summarize this article in 100 words" prompt, (b) the "summarize this podcast transcript in 5 bullet points" prompt, and (c) the "summarize this meeting transcript in 3 action items" prompt. The 3 prompts are the part I use weekly, and the weekly is the part that makes the list worth having.11. The "code this thing" list (weekly use)The list is 4 prompts for the 4 kinds of "I need to write this small piece of code" tasks I have. The 4 prompts are: (a) the "write a bash script that does [X]" prompt (the one from [the 5-line bash post), (b) the "write a Python function that does [X]" prompt, (c) the "write a SQL query that does [X]" prompt, and (d) the "write a regex that matches [X]" prompt. The 4 prompts are the part I use weekly, and the weekly is the part that makes the list worth having.12. The "translate this content" list (rarely use)The list is 2 prompts for the 2 kinds of "I need to translate this" tasks I have. The 2 prompts are: (a) the "translate this from English to [language]" prompt, and (b) the "rewrite this in [language] without losing the meaning" prompt. The 2 prompts are the part I use maybe once a month, and the once-a-month is the part that still makes the list worth having.The 5 I am embarrassed to admit I haveI lied. The 12 lists are the part I described, and the 12 lists are the part I have. The 5 more are the part I am embarrassed about, and the embarrassed is the part I want to be honest about, because the honest is the part this post is about.The 5 more are:The "rate this output" list. 3 prompts for "rate this AI output on a 1-5 scale for [criteria]." The list is the part I use when I am not sure if the output is good enough. The not-sure is the part I am embarrassed about, because the not-sure is the part that means I do not trust my own judgment. The trust-my-own-judgment is the part I have learned to want, but the want is the part I have not fully achieved.The "explain why this is bad" list. 2 prompts for "explain why this output is not as good as it could be." The list is the part I use when I want a second opinion. The second-opinion is the part I am embarrassed about, because the second-opinion is the part I should be able to give myself. The give-myself is the part I have learned to work on.The "translate this from AI to human" list. 2 prompts for "rewrite this AI output so it sounds like a real person wrote it." The list is the part I use when the output is clearly AI-generated, and the AI-generated is the part I am embarrassed about producing in the first place. The first-place is the part I have learned to fix at the prompt level, not the post-process level. The fix-at-the-source is the part I have learned is the right way.The "make this sound more like me" list. 2 prompts for "rewrite this in [my voice profile]." The list is the part I use when I am writing a post and the output is not quite my voice. The not-quite is the part I am embarrassed about, because the not-quite is the part I should be able to fix in my own head. The own-head is the part I have learned to want, and the want is the part I have not fully achieved.The "what is wrong with this prompt" list. 1 prompt for "review this prompt and tell me what is wrong with it." The list is the part I use when I have written a prompt that is not working, and the not-working is the part I am embarrassed about, because the not-working is the part I should be able to debug myself. The debug-myself is the part I have learned to do, and the do is the part I have not fully achieved.The 5 lists are the part I have, the 5 lists are the part I use, and the 5 lists are the part I have learned to be honest about. The honest is the part I want this post to be, and the post is the part I want to be the closing of the "I have a list of lists" topic that I have been promising to write about for a year.How I organize the listsThe 12 lists (and the 5 more) are organized in a single notes file. The notes file is in Obsidian. The notes file has a top-level header for each list. The lists are the part I open in the morning, and the morning is the part I pick the right prompt for the task. The task is the part I am about to do, and the about-to-do is the part I have learned to start with the right prompt, because the right prompt is the part that produces the right output. The right output is the part I want, and the want is the part I have learned to want, and the want is the part this post is about.The notes file is also the part I version-control via Git. The version-control is the part I have learned to want, because the version-control is the part that lets me revert a bad edit. The bad-edit is the part I have had, and the have is the part I have learned to recover from, and the recover is the part Git is for. The Git is the part I have, and the have is the part I want this post to be the part of my workflow that uses the lists well.What I would tell past-meIf I could go back and tell past-me one thing, it would be: the lists are the part that compound. The lists are the part I have been building for 3 years, and the 3 years is the part I have a 12-list system that I use every day. The every-day is the part I want, and the want is the part I would not have without the 3 years of compounding. The compounding is the part I would start on day one, and the day one is the part I would not skip. The skipping is the part I have learned to not do, and the not-doing is the part I want to recommend to anyone who is reading this post.I would also tell past-me: the lists are not magic. The lists are the part that are just prompts. The prompts are the part I have to update when the model changes. The model changes are the part that happen every 2-3 months, and the 2-3 months is the part I have to re-test every prompt in the list, and the re-test is the part I have learned to do, and the do is the part I want to recommend. The recommend is the part this post is for.Who this is forAnyone who is starting to build their own prompt list and is wondering what other people's lists look like. Anyone who has a notes file full of prompts and is wondering if the notes file is the right way to organize them. Anyone who wants a starting point for their own list. The 12 lists are the starting point, and the starting point is the part I want to provide, and the provide is the part this post is for.The honest partThe "12 lists" number is the part I want to be honest about. The number is the part that I have built up over 3 years, and the 3 years is the part I want to be honest about. The honest is the part I want this post to be, and the post is the part I want to close the "I have a list of lists" topic that I have been promising to write about for a year.The honest part also includes: the lists are not for sale. The lists are the part I have in a notes file, and the notes file is the part I do not share in a Gumroad pack. The Gumroad pack has different content (the [Blog Writing Factory is the part that has the prompt patterns, not the lists). The lists are the part I keep for myself, and the keep-for-myself is the part I want to be honest about. The honest is the part I have learned to want, and the want is the part this post is about.Want the patterns, not the lists?The Blog Writing Factory has the prompt patterns that go into lists like the 12 I have. The pack is the part that gives you the 5 patterns I use most: the "do this 1 thing" pattern, the "debug this output" pattern, the "ship this thing" pattern, the "research this topic" pattern, and the "decide between options" pattern. The 5 patterns are the part that go into a list, and the list is the part you build yourself. The build-yourself is the part the pack is for.Buy on Gumroad. $14
Send me the rough edges if you try it. I read every message.