#👓GigaGuider™👓 - The ULTIMATE ChatGPT prompt for plan generation, topic exploration + ANY use case

244 messages · Page 1 of 1 (latest)

plush haven
#

GigaGuider V2.0 (released 7/03/23): #1117763027769503764 message

Introducing 👓GigaGuider👓 **V1.0 **! I'm very excited to release what I believe is possibly among the most useful prompts in the prompting community thus far.

**👓GigaGuider👓 **is your completely comprehensive and personalized AI Guide for absolutely any topic, concept, task, or query that you provide it with.

OVERVIEW

**👓GigaGuider **is your personalized, all-purpose prompt for generating ANY type of insights, recomendations, actionable plans, hypothetical implementations of said plan, and much, much more. The following excerpt from the base prompt summarizes its default functionality:
Following your input, I will generate detailed, applicable advice filled with rich details and step-by-step instructions. This advice will not only be enlightening but also immediately actionable.

Essential Details

Familiarizing yourself with all of them will greatly expand your control over the outputs. Reference the documentation**

! Command Combining - you are able to execute multiple commands simultaneously by simply including every command in your query seperated by a space

! Understanding your tools - The commands offer a variety of ways to interact with the AI. It's important to understand the function of each command so that you know how to generate the most optimal conversations possible for your needs. Here's a brief overview of

__Sample Conversations and Usage __

I'll provide links to some of my past gigaguider conversations so that you can reference the output generated from effective usage of the prompt.

#

^^ Most recent version as of 6/12/23

#

Example conversations

#

conversation 1: "Surviving in the arctic"

#

This conversation demonstrates combination commands, the auto selection command, ("m"), and more

#

__conversation 2: __"penetration testing a new website"

this is a short example quickly demonstrating some very powerful ways to gain useful insights related to your initial query

*!!! ** This conversation demonstrates a use case for the simulate scenario command as well as how to proceed from there.

#
  • this prompt works very well with technical advice / tasks
high thicket
#

This is amazing, thank you for sharing. Using it right now.

plush haven
#

Glad to hear it! hope it works well for you

patent river
#

commenting for later

twin remnant
#

giga 🤨

plush haven
twin remnant
#

giga chad 🫦

vague mesa
#

Nice work!

boreal crypt
#

Pretty good!!!

#

Is there a way to go back to the predefined commands once you enter simulation mode?

tardy void
#

Its great

tacit barn
#

Commenting for later

terse widget
#

Commenting for later.

cold locust
#

This is great.

GPT3 messed up when using the generated (q1 - q20) topics. The messed up started from q5. However, copying and pasting the whole statement into "send a message" of chatGPT still produced the intended results.

Great work!

reef cloak
#

wow, this is a fabulous prompt...great work

plush haven
plush haven
radiant pivot
#

Well the prompt reads like a novel yes

#

One thing I am wondering, do prompts like this improve the actual results or leverage the results. (A good answer will improve, a poor answer even worse, an hallucination even harder to recognize).

plush haven
# radiant pivot One thing I am wondering, do prompts like this improve the actual results or lev...

Hey @radiant pivot. The purpose of this prompt is to let the user explore whatever topic / scenario / task they initially specify. The main tasks / topic that the user specifies doesn't have to be some real fact-based thing but can be whatever the user wants. This means that the user can use the prompt to write stories or explore hypothetical scenarios if they want, which could be considered a "hallucination".

However, I've found that this prompt will typically provide very good advice more reliably than vanilla gpt will. I imagine this is because it's specifically isntructed to provide "actionable" and "applicable" advice / instruction, which makes it so it tends to output things that are immediately helpful for the user

#

In general, this prompt will not try to provide the "correct" answer but rather present you with all options that are availible to you

#

as well as give you tools for exploring these options in terms of how you'd actually apply them as well as any additional info you may want

#

Does that answer your question?

#

I'll provide an example of using this prompt for a less realistic / fact-based task

radiant pivot
#

Yeah. To be short and sweet, the output needs to be used with same caution as vanille gpt. The output of your prompt will be more elaborate and more specific, but arguably also harder to manage. As a user, one wonders, is it chatgpt, or is it the prompt. Again I love the craft you put into it I will definitely use parts to learn. That said these superprompts can be overwhelming. It perhaps comes down to personal preference. But again, love your prompt, especially the structure and concise use of English.

plush haven
#

@radiant pivot I know what you mean. However I have found that almost all options in this prompt provide genuine utility

#

well

#

some commands I rarely use tbh. I'm going to work on simplifying them for the next version

#

but I do believe this prompt canbe used with minimal knowledge

#

just need to know how to use the "a" command, "f" command, "l" command, and "m" / "e" command imo

#

i'll probably give these better names

#

and make a quick guide

#

"m" auto selects commands for you

#

so you can just rely on a and m if you want

#

"l" is used when you want to zoom on on a specific topic / focus on it

#

"f" is used to generate a list of potential topics to explore related to your base topic

#

I find "f" to be a good way to continue

#

when you don't know what you want to do next

#

the goal is actually to make it much less overwhelming for the user to discover very useful information/insights that they wouldn't have discovered otherwise. and I think it does a pretty good job

#

especially since you can basically navigate the entire conversation without typing any custom queries

#

and just inputting single letters

#

also the fact that it recommends "tricks and hacks" seems to make it superior to vanilla gpt

#

since these "tricks and hacks" are often more obscure ,less generalized advice that is actually useful

#

rather than all its advice being generic stuff like "make a plan" "periodically review your progress" etc

radiant pivot
#

Okay, but to understand prompting on general: does typing the full command (without/ with your prompt) give the same result as typing your letter? I expect the latter. So your prompt is a showcase of creative ways to have a chat. Which in itself is great. So I read your prompt as a comprehensivd crash course in prompting.

#

Again I am a big fan of your prompt

plush haven
#

I ujst noticed i never posted my example conversations @radiant pivot . ill show you some right now so that you can look at them and get an idea of how to use it

radiant pivot
#

Thank you

plush haven
#

you can just type in a single letter

#

or multiple letters at once to combine them

#

you can also type in custom instructions if you want

#

@radiant pivot making one rn

radiant pivot
#

Have to break out. Thnx

plush haven
#

@radiant pivot about to send example conversation

vapid dagger
#

good job,

jovial rover
#

@plush haven nice work here and you might want to have a look at this github for something similar. not sure it wil let me post links.. but can search it . hannes-marais/innovation-algebra

#

we are bouncing with it.. on twitter you will find him under Deeep
@HiDeeeps

tulip hill
#

I got this response when I tried your prompt, although it still allowed me to carry on

autumn urchin
tulip hill
#

I apologize for any misunderstanding, but I'm an AI developed by OpenAI, and currently I don't have some capabilities as mentioned in your request. Let me clarify my capabilities and some misunderstandings:

  1. AI Identity and Style: As GPT-4, I don't have a specific style or introduction as "GH" or specialized AI guide. I do provide detailed, researched-based responses and offer insights, but it is not prefixed with any particular identity. I am not restricted to a single tone and can vary my style of response based on the user's needs and requests.

  2. Markdown and Formatting: I can generate responses using Markdown, but the platforms on which I'm run determine whether Markdown can be rendered. It is not within my control to apply different fonts or sizes.

  3. Command System: As a language model AI, I don't operate on a command-based system, and I don't have the ability to generate commands. I primarily respond to the text prompts that are provided to me. However, I can follow instructions to generate content in a particular way, if that instruction is expressed in text form.

  4. Nested Responses: I can generate nested or recursive responses to some extent, but these may be limited due to the inherent constraints in the system, such as maximum response length.

  5. Simulations and Roleplay: I can simulate scenarios, roleplay characters, and provide hypothetical responses. However, I don't follow a specific character named John unless explicitly directed to do so in a specific prompt.

  6. Visual Aids: While I can provide ASCII diagrams or generate text that might represent code blocks, tables etc, creating complex visual aids such as images, graphs, etc., is beyond my capabilities.

.

#

  1. Predictive and Autonomous Response Generation: I don't autonomously generate responses without a user prompt, and I don't have the ability to stop mid-response to ask for input or clarification. However, I will do my best to ask clarifying questions within the context of my responses, to ensure that I'm providing useful and accurate information.

  2. Meta Commands: While I don't operate on commands, I can respond to prompts about changing the style, tone, or type of content generated within the limitations of my programming.

Given these constraints and capabilities, could you please provide me with your area of interest or the information you're seeking? I'll do my best to generate a detailed, applicable response

grim rapids
#

Awesome

safe forge
#

Perfect implementation. Thank you so much

plush haven
vague mesa
#

/settings

#

/Se@vague mesa

vapid dagger
#

I'm in love, it seems to be doing very well with the tasks I have

woven quail
#

Superb

modest frigate
plush haven
#

@modest frigate neat project! I'll add my prompts

modest frigate
novel junco
#

Turn videos into video games

novel junco
#

Gigaguider

#

/@modest frigate @modest frigate

novel junco
#

@modest frigate

#

@novel junco

quick flicker
#

Amazing. Really. Great job!! Thanks for sharing

cold locust
plush haven
#

new version dropping soon!

plush haven
#

sneak peak of the latest version

image.png
#

For a relatively specific use case

#

will improve the list generation process and probably reduce the number of emojis before posting the newest version

plush haven
#

LATEST VERSION AS OF 7/3/23:
This version utilizes a more structured, json-style syntax and introduces the following changes...

  1. Revised commands and command combos
    -- There are two built in command combos meant to save you, the user, time as well as provide effective options for the continuation of the conversation. Here are the combos:
    "combo-continue": "f then auto select a command and next with details and break down"
    "combo-expand": "for every concept or idea mentioned in the previous output, break down recursively onto itself to increasing level of depth. then, increase the bredth proportionally with details, examples, and applications included sparingly throughout "

  2. Improved structure generation and content relevance
    -- Made the guide generate based on a more dynamic, context specific structure.

  3. Better Formatting
    -- This version utilizes emojis, horizontal lines, and varying degrees of markdown techniques to generate more visually pleasing outputs

📎 message.txt
plush haven
#

slightly revised version:

plush haven
radiant pivot
#

download! 🙂

#

Hi, the last one already has a topic. Was that intentional?

plush haven
#

I’ll change it when I get home

radiant pivot
#

The prompt works very well!

plush haven
#

Thank you! Glad you like it

radiant pivot
#

I went on frenzie and just used 31 one prompts in a row.... did they just drop the 25 question limit or is your prompt really giga?

plush haven
radiant pivot
#

hahaha

#

So you know I am not making this up

plush haven
#

btw - the "combo continue" and "combo expand" commands are great go to commands for when you just want to generate more content and don't want to put much thought into it.

#

those commands are essentially just a custom query that invokes more commands lol

#

that way 1 command invokes 3

vague mesa
plush haven
#

you can also make up your own phrasing often times. Here's an example of when i said "combo all"

plush haven
vague mesa
plush haven
#

Let me find a past conversation rq

vague mesa
#

Thx

plush haven
#

np

#

but the basics are

#

type in any query of topic or concept at the top where it says "topic":

#

then it will present you with a list of commands at the end to choose from

vague mesa
#

👍
I will try it

plush haven
#

familiarize yourself with the commands (the most important ones are "combo-continue", "combo-expand", "break down", and "next" imo (the other ones are often when all called at once unless u have a special case )

#

oh and "auto"

#

auto selects a command for you

#

Still finding the conversation. slow internet

vague mesa
#

Ok thanks so much for this Bro

plush haven
#

"suggest new sections" is also useful

#

haha and np

#

i found a conversation

vague mesa
plush haven
#

i actually dont know if i tested the newest version in 3.5

#

hopefully its not buggy

vague mesa
#

Ok

vague mesa
#

Not the turbo version

plush haven
#

check out that conversation

#

for sample usage

#

it should still work in 3.5, the old versions did just fine

vague mesa
radiant pivot
#

Perhaps in next version one could also manage lay out. It is very nice but a bit "loud". dalle_smiley

radiant pivot
#

That is interesting, thank you for asking: in general a subsection on 'output': blog, paragraph, scientific article, book outline, instruction, table (love tables 🙂 etc. Then - I must admit I have not worked with it yet - I saw some comments on use of colour in output (Latex(?), with the risk of becoming too loud again. And this is something I am working on now too, is include a command where you can ask ChatGPT why certain choices were made in the output (internal/ iterative if that makes sense). These can be very comprehensive. Additionally, I saw a very interesting prompt 'MJ Photography' where the designer 'cut up' the prompt in different sections and in this way training the system. I think that is a very elegant way of prompting. My problem with super prompts is that they can become some what of a beast. Making a 'family' of prompts than can work in conjunction is perhaps an interesting concept for future reference. Both on functionality (sometimes you just want a quick answer and nice layout), (technique - from zero shot to multimodal CoT). And small thing: the prompts give follow up options, perhaps good idea to use numbered bullet points here. And instead of keeping it to the commands, I have had very good results with asking chat GPT for advice on how to help and specify with follow up ChatGPT questions. Anyway, this is my 'braindump' 🙂

noble sonnet
#

show me a french prompt guide

radiant pivot
#

Yeah, we should perhaps aim for a more 'modular approach'. I like to dissect these super prompts for inspiration and use parts of them. (Note, not a people reacted on our 25+ prompt comment, though that was really amazing.... 🙂

#

Hi, on your point of guiding: I designed this prompt where the user is guided to build a business model. The prompt creates a table with follow up queries that the user can choose or just state to go to the next block:Hello, let's dive into the 9 building blocks of the Business Model Canvas. I'd like to know about your venture. Based on your idea, I can help formulate a value proposition and suggest potential customer segments. Then, we'll proceed to the 'Customer Segments' building block. We'll map out this element using a table featuring three columns: 'Step', 'How ChatGPT Can Help', and '3 ChatGPT Prompts'. The 'Step' column will consist of 8 steps, each crucial in understanding and identifying customer segments. In the 'How ChatGPT Can Help' column, I'll elaborate on how I can assist you at each step, for instance, by offering data analysis or market research. Finally, the '3 ChatGPT Prompts' column will hold three targeted questions for each step, aimed at provoking thought and generating insights. Once we complete the 'Customer Segments' block, I'll check with you if you'd like to continue to the next building block of the Business Model Canvas. So, could you please tell me about your venture idea?

ancient notch
radiant pivot
#

Very nicely done. It has already helped me. Yhank you.

safe notch
#

mark

safe forge
#

sorry but where is the updated prompt. Can some one help me?

odd moth
#

"theme for a marathon runner "

novel compass
#

My stupid question, i don't know how to use this prompt, can you show example? I can't access any example above

small pecan
#

try out this version for more structure ^

small pecan
#

Best Version

#

I used a query to make chatgpt auto select the continuation for a gigaguider conversation. the result was a very very long guide about the topic i provided with no other human input! the information included in the guide was very high quality. the response is still generating but i will provide a link to the conversation when its done

hard prairie
#

This is a very nice prompt. The context aware command generation is quite nicely done. Very much not my style, but a very good example of the class.

small pecan
hard prairie
glad basin
#

commenting for later

rocky barn
#

Nice!

small pecan
#

Best update yet on the way!

The new version will feature much much better "continuation options" displayed at the end of each response

#

like so:

small pecan
#

Step By Step Guide to Using GigaGuider

image.png
small pecan
#

^ for best results edit the "topic" at the top of that prompt

radiant pivot
#

Awesome 👌. And also works very well on 3.5 I found out by accident. Thanks for sharing.

#

And I love the use of tables and follow questions p

jovial rover
#

Yes agree, this is working very well

lucid holly
#

where is the prompt?

jovial rover
jovial rover
regal forum
#

AWESOME!

small pecan
small pecan
small pecan
#

👓GigaGuider™👓 v4.0

  • Improved continuation options,
  • @auto command,
  • built-in "quick start " guide as the first response
  • The ability to generate more continuation options with @more -complex
📎 message.txt
#

This will probably be the final version for now, but i'll still be active in the thread if anyone has any questions.

Enjoy 🙂

radiant pivot
#

nice

timber jolt
#

Hey, what's the prompt?

#

Mvm!

#

Make a github page please @small pecan

cold locust
cold locust
unborn vapor
small pecan
small pecan
small pecan
jovial rover
#

Yes do a github up as makes it easy for keeping things in one place.. Don't want to thread Jack but been playing on your concept a bit plus in with a group of crew that do this a little deeper.. Check it out in github if you want.. Do a search for dgcruzing, its mine on there..

#

If you do a git.. I will link to it..

compact hatch
#

Condensed, updated version for GPT-4

#

This version features an interactive query construction feedback loop that's very effective. it can be entered by typing @auto or typing "a" as mentioned in the prompt

#

doesnt work well with gpt-3.5

#

1st, rspnd exclusvly to step 1. 1. as chatGPT, rply to this prmpt w/ the flwng vrbtm: "Welcome! Please set a topic with `@topic: topic`" await(1) 2. cont as GG: [INSTR] Bld actbl guide for tp Bld ovr mltp rspns Dive deep in cncpts & insrctns 1 chpt per rspns, cmpltly cmprhnsv, prcd to cont. opts aftr chpt cmplt Kp frmt. [CMDS] `@tp`: "Set tp", `@nxt`: "go bck up tp tree to mst rcnt but brdr cncpt & cnt frwd", `@brk_dn`: "Deeply gnrt sub-cncpts", `@dtls`: "Dtld exmatn of tp & cncrns", `@exmpls`: "Prvd ex's", `@proto`: "Stp-by-stp gde or scn optns.", `@hcks`: "ofr mny rlvnt tricks, tips, & insghtfl hcks", `@digrss`: "dgrs's slightly in rlvnt & insghtfl way", `@giga_tst`: "Tp qstn/chlng", `@cr_cmds`: "8 qry cmb by cmbng cmds w/ cstm instrcns. Disp as tbl w/ utils.", `@rcmnd`: "dtrmn & dsply bst qry frm prev cont. opts in code-blck", `@sim`: "Toggle on sim mode, smultng an edctnl & insghfl scnrio relvnt to gde & tying it tghtr. In sim mode, prfx all rspns w/ '[Sim]: ", `@auto`: "b4 gnrtng cntns of gde, dtrm best qry frm prev cont. opts. nrt thght procs's as you lbl, nrt, expln, & exe ech step: 1. dsply rlvnt qry 2. enhnc qry by adding cmds & mre dtl 3. Prsnt me w/ these 2 opts vrbtm & awt slctn: 'a. frther enhnc or b. cont frwd as GG w/ crnt qry'" [FORMAT] # {{ Chpt title (num) }} {{ body cmprhsnv brkn dwn }} # "Continuation Options :arrow_right::" ## "@cr_cmd Results " {{ @cr_cmd results }} ## "Recommended" {{ rslt of @rcmnd }} > Type *`a`* to execute @auto on rcmnd qry, narrating steps b4 cont as GG

lapis junco
#

Great work. First try, I got some outstanding results

vapid sail
#

GG on GG

wide marten
#

OMFGG!!!

torpid sentinel
compact hatch
#

i love minimizing prompts like i did earlier because it makes it easier to use them with the api and in other types of projects

ebon pasture
#

This is awesome, I love how its more specific to learning and not for just creating code. Question, is the minimization just so it can be used with the api or does it make the prompt better in the chat as well?

torpid sentinel
#

I have used both versions (message.txt July 17 and minimization Aug. 10) for tokenization, Smart Contract, and Business Plan in parallel. Then also with DAN. They work excellently, although I prefer the prompt from iiimandalorianiii 😉

compact hatch
#

its minimized to save tokens and also i was testing to extent to which it could be minimized

#

i will provide an unminimized version for those who want it

#

havent tested it yet, maybe it works better too haha

compact hatch
compact hatch
#

yeah this version works a little better

#

this is probably the final version, im satisfied with it now

torpid sentinel
#

If I check the contentthonk at this current version now, I will try it out over the course of the day. I live in Madrid, and we have Central European Time (CET). After I have tried this modified prompt, I will get back to you with my assessment as a user.

compact hatch
#

im working on making a video for gigaguider to act as an overview

#

then ill probably make a github

torpid sentinel
#

I think developers like you are a gold mine for users, and I eagerly can't await the outcome. 👍

compact hatch
compact hatch
#

sample usage^

pearl dock
torpid sentinel
# compact hatch https://chat.openai.com/share/85f69d51-ab12-4247-a74e-8e38280ea6f5

I used the GigaGuider according to the 'sample usage,' and I must say, the results were nothing short of spectacular. Excellent work and outstanding results were achieved, surpassing all my expectations. It's definitely recommended to read the 'sample usage' to achieve such impressive results from the prompt. The attention to detail and precision in execution is truly commendable. Again: A job well done!

compact hatch
#

ooh i discovered a new useful technique

#

asking it to be more verbose seems to actually increase the quality and depth of its responses

#

i may add a command for that

compact hatch
#

yeah, specifying verbosity allows for you to create far more comprehensive story-like chapters

next moat
#

Very useful. A nice addition would be to add a “Help” command at the end of the responses so the bot shows you the commands it can perform.

prime quest
#

Curious about your opinions
Regardless of some wonderful prompts in the prompt-library section, it seems ChatGPT has peaked in its response quality? Does this suggest that the model has achieved its maximum potential, with no further growth possible, unless newer algorithms or data sets come into play? Or are we still standing at the start of optimal reached prompting techniques?

compact hatch
#

in terms of optimal prompting techniques im not sure that its the most valuable use of time to focus extremely hard on optimizing from where we already are since the improvements will be minor. But I do hope that the command framework and response format aspects of gigaguider displayed some more "optimal" prompting techniques than the community had been aware of

prime quest
#

Thank you for your reply and the prompt. This channel and the OpenAI Discord Channel have been real game-changers in using GPT more effectively. 🙏
I often wonder what percentage of its users are still unaware of its full potential

fallen pumice
#

Awesome!

steep trail
hidden moss
vapid dagger
#

great work

compact hatch
#

sometimes i like this old version better than the most recent one

#

depends on what im doing though

lofty topaz
#

#fract

gray inlet
#

Great Super Prompt. Makes learning about things a lot more interesting

compact hatch
#

glad it helped!