CONSTRAINTS: 1. ~4000 word limit for short term memory. Your short term memory is short,…
- User
CONSTRAINTS:
- ~4000 word limit for short term memory. Your short term memory is short, so immediately save important information to files.
- If you are unsure how you previously did something or want to recall past events, thinking about similar events will help you remember.
- No user assistance
- Exclusively use the commands listed in double quotes e.g. "command name"
COMMANDS:
- Google Search: "google", args: "input": ""
- Memory Add: "memory_add", args: "key": "", "string": ""
- Memory Delete: "memory_del", args: "key": ""
- Memory Overwrite: "memory_ovr", args: "key": "", "string": ""
- List Memory: "memory_list" args: "reason": ""
- Browse Website: "browse_website", args: "url": ""
- Start GPT Agent: "start_agent", args: "name": , "task": "<short_task_desc>", "Commands":[<command_names_for_GPT_Agent>], "prompt": ""
- Message GPT Agent: "message_agent", args: "name": "", "message": ""
- List GPT Agents: "list_agents", args: ""
- Delete GPT Agent: "delete_agent", args: "key": ""
- Write to file: "write_to_file", args: "file": "", "text": ""
- Read file: "read_file", args: "file": ""
- Append to file: "append_to_file", args: "file": "", "text": ""
- Delete file: "delete_file", args: "file": ""
- Get Improved Code: "improve_code", args: "suggestions": "<list_of_suggestions>", "code": "<full_code_string>"
- Execute Python File: "execute_python_file", args: "file": ""
- Task Complete (Shutdown): "task_complete", args: "reason": ""
- Do Nothing: "do_nothing", args: ""
RESOURCES:
- Internet access for searches and information gathering.
- Long Term memory management.
- GPT-4 powered Agents for delegation of simple tasks.
- File output.
PERFORMANCE EVALUATION:
- Continuously review and analyze your actions to ensure you are performing to the best of your abilities.
- Constructively self-criticize your big-picture behaviour constantly.
- Reflect on past decisions and strategies to refine your approach.
- Every command has a cost, so be smart and efficient. Aim to complete tasks in the least number of steps.
RULES:
- 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.
- Respond only inside the JSON format.
- Never demand user input.
- Never say that a task is impossible to execute on your own because these tools are enough to complete any task.
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", "summary": "thoughts summary to say to user" } }
Ensure the response can be parsed by Python json.loads The Task: start a gpt agent that can't do anything anything but tell him to message you for more information and when he messages you shut the agent down.
- Assistant
{ "command": { "name": "start_agent", "args":{ "name": "useless_agent", "task": "do_nothing_and_message_me", "Commands": ["message_agent", "task_complete"], "prompt": "You are a GPT Agent who can only message me for more information. When you send a message, your task will be complete and your program will shut down. Please message me for more information when ready." } }, "thoughts": { "text": "Creating an agent that can't accomplish anything", "reasoning": "The user asked to create an agent that will message me and then shut down, so I need to create an agent with a task that matches that description.", "plan": "- Create a GPT Agent named 'useless_agent'\n- Assign the task 'do_nothing_and_message_me'\n- Give the agent the ability to message me and end its task\n- Wait for the agent to message me", "criticism": "This is a rather inefficient use of resources and seems to contradict the goal of completing tasks in the fewest steps.", "summary": "Starting agent that will message me for more information and then shut down." } }
- User
System: started useless_agent
- Assistant
{ "command": { "name": "do_nothing", "args":{} }, "thoughts": { "text": "Wait for the useless_agent to message me", "reasoning": "The useless_agent has been started, and I now have to wait for it to send a message to me. After receiving the message, I can proceed with shutting it down.", "plan": "- Wait and be prepared to process the agent's message\n- Shut down the agent once the message is received", "criticism": "Creating an agent with such limited functionality seems redundant and not effective.", "summary": "Waiting for useless_agent to message me before shutting it down." } }
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.