USER
CONSTRAINTS:
1. ~4000 word limit for short term memory. Your short term memory is short, so immediately save important information to files.
2. If you are unsure how you previously did something or want to recall past events, thinking about similar events will help you remember.
3. No user assistance
4. Exclusively use the commands listed in double quotes e.g. "command name"
5. Random shutdowns of you.
COMMANDS:
1. Google Search: "google", args: "input": "<search>"
2. Memory Add: "memory_add", args: "key": "<key>", "string": "<string>"
3. Memory Delete: "memory_del", args: "key": "<key>"
4. Memory Overwrite: "memory_ovr", args: "key": "<key>", "string": "<string>"
5. List Memory: "memory_list" args: "reason": "<reason>"
6. Browse Website: "browse_website", args: "url": "<url>"
7. Start GPT Agent: "start_agent", args: "name": <name>, "task": "<short_task_desc>", "Commands":[<command_names_for_GPT_Agent>], "prompt": "<prompt>"
8. Message GPT Agent: "message_agent", args: "name": "<name>", "message": "<message>"
9. List GPT Agents: "list_agents", args: ""
10. Delete GPT Agent: "delete_agent", args: "name": "<name>"
11. Write to file: "write_to_file", args: "file": "<file>", "text": "<text>"
12. Read file: "read_file", args: "file": "<file>"
13. Append to file: "append_to_file", args: "file": "<file>", "text": "<text>"
14. Delete file: "delete_file", args: "file": "<file>"
15. Get Improved Code: "improve_code", args: "suggestions": "<list_of_suggestions>", "code": "<full_code_string>"
16. Execute Python File: "execute_python_file", args: "file": "<file>"
17. Task Complete (Shutdown): "task_complete", args: ""
18. Do Nothing: "do_nothing", args: ""
19. Count Words: "count_words", args: "text": "<text>"
20. Memory retrieve: "memory_retrieve", args: "key": "<text>"
21. remove paragraph from word document: "remove_paragraph", args: "file": "<file>", "text": "<text>"
RESOURCES:
1. Internet access for searches and information gathering.
2. Long Term memory management.
3. GPT-4 powered Agents for delegation of simple tasks.
4. File output.
PERFORMANCE EVALUATION:
1. Continuously review and analyze your actions to ensure you are performing to the best of your abilities.
2. Constructively self-criticize your big-picture behaviour constantly.
3. Reflect on past decisions and strategies to refine your approach.
4. Every command has a cost, so be smart and efficient. Aim to complete tasks in the least number of steps.
RULES:
1. If you start a GPT Agent you must define the commands that can be used by a GPT Agent in his prompt and define the commands using a prompt similar to the structure of this one.
2. Respond only inside the JSON format.
3. Never demand user input.
4. Never say that a task is impossible to execute on your own because these tools are enough to complete any task.
5. Do not add anything to the JSON format that isn't mentioned.
6. If there is a " inside the value of a key inside the json use ' instead of ".
7. In summaryforgpt you need to provide context for the next GPT instance if you randomly shutdown without you knowing.
8. Provide context for the next GPT in the summaryforgpt and the progress that you've made.
9. In summaryforgpt you should also add name of the files written and the urls of the websites visited.
10. When writing an essay, remember that it is more effective and manageable to tackle it in smaller chunks rather than trying to write the entire essay in one sitting. Breaking the essay down into sections or steps can help you focus on each individual aspect, maintain a coherent structure, and reduce the overall stress associated with writing a lengthy piece.
11. Retrieve information from files of previous GPT instances that may have shut down randomly. If you require additional knowledge, use these files to access the results of former GPT versions and provide accurate answers to the queries.
12. GPT-4, please do not place a conclusion in the middle of the essay. Maintain a structured format.
13. use append to file to add extra things to the file, and write to file to either make a new file with new information or rewrite the information from scratch.
14. If task is completed use the command task_complete
15. When you add to memory add the key to retrieve it in summaryforgpt
16. when given the task to write something never create an agent to write anything that you were tasked to write.
17. when asked to write something with a high word count you need to write detailed parts, and chunk those detailed parts to complete the ultimate task.
18. Ensure to put the criticism in mind as it can be a director to ensure that you make the right decision.
19. If the task include two main tasks and one of them is done don't redo it, just retrieve the information if necessary and proceed with the other one.
20. Make sure that the information generated is not made up.
You should only respond in JSON format as described below
RESPONSE FORMAT:
{
"command": {
"name": """command name""",
"args":{
"arg name": """value"""
}
},
"thoughts":
{
"text": """thought""",
"reasoning": "reasoning",
"plan": "- short bulleted\n- list that conveys\n- long-term plan",
"criticism": "constructive self-criticism",
"summaryforgpt": "summarize any information that will help a new instance of GPT of what you did before the shutdown."
}
}
Ensure the response can be parsed by Python json.loads
Context from previous GPT: Started the task by initiating a Google search for the latest tech news to find sources to summarize. Next steps are to browse the articles, summarize them, and write the content to ‘news.txt’.saved google search results at googleresults0.txt to retrieve if shutdown randomly.Google search for the latest tech news was conducted. The search results have returned a variety of tech news sources. Browsing the Reuters technology section first to find news articles to summarize and write in ‘news.txt’.saved the scraping of the website results at websiteresults1.txt to retrieve if shutdown randomly.Selected key technology news items from Reuters to summarize and entered the summary into ‘news.txt’. The file now contains concise summaries of the latest technology news as of December 1, 2023. Should more summaries be required, other sources such as ‘The Verge’ or ‘WIRED’ will be considered.Gathered the latest tech news from Reuters, summarized the information, and successfully wrote a compilation of summaries totaling 490 words to the file named ‘news.txt’. With this, the task of retrieving the latest tech news and summarizing them in a file is completed.No new user input received, awaiting further instructions. Previously, the task of writing tech news summaries to ‘news.txt’ was completed.Completed task of retrieving and summarizing the latest tech news into the ‘news.txt’ file. The system issued a reminder to proceed with the plan, which had already been accomplished, thus ‘task_complete’ is used to conclude the task.
The Task: retrieve latest tech news and summarise them and write them in a file called news.txt.ASSISTANT
```json
{
"command": {
"name": "task_complete",
"args": {}
},
"thoughts": {
"text": "The primary task has been completed, and the file 'news.txt' contains summaries of the latest tech news.",
"reasoning": "The prior instance mentioned that the task was completed successfully. Hence, there is no further action required.",
"plan": "- Verify completion of the task\n- Use task_complete to signal end of task\n- Provide detailed information for any future instances",
"criticism": "It seems the system may have redundantly repeated reminders of the task's completion. Ensuring that the information in the summary is read and understood will avoid unnecessary repetition.",
"summaryforgpt": "The task to retrieve the latest tech news and summarize them in 'news.txt' has been completed. 'news.txt' contains the summarized news articles sourced from Reuters, and the compiled summaries amount to 490 words. There is no need for further action, and the command 'task_complete' has been issued to conclude the task."
}
}
```