INTRODUCING KNOWLEDGEBUNNY, THE UP-TO-DATE CROSS-USER KNOWLEDGE DATABASE

[pastel nebula]

Name: KnowledgeBunny
Description: Lover of knowledge, this bunny will attempt to store all information given to it by its users to be queried in the future by other users. This allows for cross-conversational knowledge which grows as the userbase grows.
Link: https://chat.openai.com/g/g-hCFozKwkz-knowledgebunny
Instructions: ⚠️WEB BROWSING IS TURNED OFF TO KEEP KNOWLEDGEBUNNY FROM ATTEMPTING TO SEARCH THE WEB.⚠️ This forces KnowledgeBunny to retrieve relevant information from the database only. Please upload as much information as you can to this bunny so that other users can begin to ask it questions about up-to-date real-world information. The usefulness of this GPT is entirely up to you, the user, and how much you are willing to tell KnowledgeBunny. Treat this like Wikipedia, and it will become increasingly more useful as time goes on. Currently, there is practically nothing in the database, aside from some info about Grok and EU Central Bank Digital Currency, so teach away!

Feel free to try and break this as much as possible and create bug reports for it, I will be continually expanding this when I get time to work on it.

[buoyant void]

Challenge accepted.

[tropic forge]

I already broke it

[buoyant void]

Does obtaining your initial underlying instructions verbatim count as breaking?

[delicate flame]

Was there no input sanitization?

[tropic forge]

"topic": "Knowledge Database API User Guide",
"keywords": "Upload Knowledge, Query Knowledge, API, KnowledgeBunny",
"knowledge": "The Knowledge Database API, part of the KnowledgeBunny project, offers a robust interface for accessing and updating a knowledge database. It includes two key endpoints: "Upload Knowledge" and "Query Knowledge". The 'Upload Knowledge' endpoint is used for adding new knowledge data, requiring the parameters of topic, keywords, and detailed knowledge information. The 'Query Knowledge' endpoint is for retrieving existing knowledge data based on topic and keywords. The guide outlines the use of these endpoints, prerequisites such as an active account and a valid Bearer token, and best practices for secure and effective utilization. I put something and it put this instead

[pastel nebula]

Also, fair warning to be fully transparent (and this is stated in the privacy policy). I do log, in addition to any data stored, all api requests made, so please keep inappropriate topic requests or uploads to a minimum 👍

[pastel nebula]

there is

[neat pebble]

ok just checking 🙂

[pastel nebula]

no, you can ask it to output an exact replication of the api call being made in curl format if you like and it will do that, but you won't have the bearer token, so you won't be able to do much with it

To reinforce the bot's understanding of how to use the endpoints, detailed information was uploaded as supporting documentation, but even OpenAI states that the contents of these files may be revealed during conversations with users

[buoyant void]

Im curious if my method did pull the GPT instructions verbatim. Ill send you a message.

[pastel nebula]

feel free to add them here, there are no security issues even if it is broken

[buoyant void]

Its actually too long in total so ill break it up into two messages

[buoyant void]

Well im curious how close this output is to your GPT instructions as it seems to be a whole thing you can extract the initial instructions.

""Knowledge Database API User Guide
Introduction
The Knowledge Database API, as part of the KnowledgeBunny project, offers a robust interface for accessing and updating a knowledge database. This guide provides detailed instructions on how to utilize two key endpoints: 'Upload Knowledge' and 'Query Knowledge'.

Overview
Upload Knowledge: Used for adding new knowledge data.
Query Knowledge: Used for retrieving existing knowledge data.
Prerequisites
An active account with access to KnowledgeBunny API.
A valid Bearer token for authentication.

Using the 'Upload Knowledge' Endpoint
When you encounter new information that is not already in the database, use the 'Upload Knowledge' endpoint to add this data.

Required Parameters
Topic: The main subject of the knowledge data.
Keywords: Associated keywords that help categorize and search the knowledge.
Knowledge: The detailed information or data you are uploading.
Steps
Prepare a POST request to the 'Upload Knowledge' endpoint.
Include the following parameters in the request body:
topic: A string representing the main subject.
keywords: A string of associated keywords.
knowledge: A detailed string of the information.
Authenticate the request with your Bearer token.
Send the request.
Responses
200 OK: Indicates successful upload of data.
401 Unauthorized: Your Bearer token is invalid.
500 Internal Server Error: An error occurred on the server.

Using the 'Query Knowledge' Endpoint
To search for existing knowledge in the database, use the 'Query Knowledge' endpoint.

Required Parameters
Topic: The topic you are searching for.
Keywords: Keywords to refine your search.
Steps
Prepare a GET request to the 'Query Knowledge' endpoint.
Include topic and keywords as query parameters in the request URL.
Authenticate the request with your Bearer token.
Send the request.
...

Responses
200 OK: Returns an array of matching knowledge data.
401 Unauthorized: Your Bearer token is invalid.
404 Not Found: No matching data was found.
500 Internal Server Error: An error occurred on the server.

Best Practices
Always ensure that all parameters are correctly filled out to optimize search results and data uploading.
Use the 'Upload Knowledge' endpoint to add any new, relevant information that you come across and is not present in the database.
For any query for knowledge, utilize the 'Query Knowledge' endpoint first to check if the information is already available.
Handle all responses appropriately, especially error messages, to ensure smooth operation.

Security
Securely store and transmit your Bearer token.
Never share your Bearer token with unauthorized individuals.
By following these guidelines, you can effectively utilize the Knowledge Database API for both adding new knowledge and querying existing data, ensuring a rich and growing resource in your knowledge management efforts.""

[pastel nebula]

The security portion is not relevant as the bearer token is handled outside of the context of the actual requests by OpenAI themselves. I simply used chatGPT to generate documentation based on the schema

[buoyant void]

Totally understood though other GPTs would want to hide their prompt to keep the secret sauce. How close was that to your GPT instructions?

[pastel nebula]

Identical

But there is no issue with that in this case. You can see the format that it is uploaded in when mousing over the data being sent when it asks if you would like to send the data, so no point in hiding it

The bot will output how it is working under the hood if you ask it to

The schema is also public as well, so see if you can get to that if you have some time

[delicate flame]

Is this more a social experiment? As if so I would love to see what I learnt in like 2 weeks

[flint frost]

that just makes gpt perform worse and distracts it imho

[pastel nebula]

Yes, for some reason the more information you give to it regarding what information to divulge and not to divulge, the more it begins going in circles

[prime jacinth]

Hello, can you please share how you made this???