Automatic prompting

[Torantulino]

Background

Prompt engineering, although in English, is a bad user experience that takes skill and practice.

This addition removes the need for laborious prompt engineering from Auto-GPT, by creating a system so easy that anybody can use it effectively.

Instead of having to manually enter the ai_config information, the user can simply now type something as simple as "Make pong", and a suitable ai_config will be automatically generated.

Changes

• Adds automatic ai_config generation as the default user flow.
• Relegates manual ai_config creation to an optional "manual mode"
• Adds tests to cover automatic and manual ai_config creation
• Displays AI details to user upon creation.

Documentation

Docstrings and comments throughout

Test Plan

I have implemented unit tests to cover the new and existing functions in the setup.py file.

The prompt system prompt will likely need to be further improved going forward so that it leans on AutoGPT's skill-set.

PR Quality Checklist

• My pull request is atomic and focuses on a single change.
• I have thoroughly tested my changes with multiple different prompts.
• I have considered potential risks and mitigations for my changes.
• I have documented my changes clearly and comprehensively.
• I have not snuck in any "extra" small tweaks changes

[richbeales]

linting fail

[ntindle]

add a note to news for this

[ntindle]

And fix the linting

[ntindle]

also you mayyyyy have infinite tests so take a look at that

[Torantulino]

CI Tests should be fixed in #2904

[Pwuts]

This is definitely going to be good for UX. :)

[Torantulino]

Would be great to see some tests and prompt tweaks on the system prompt here.

It's not perfect yet, as you can see in the example it includes steps that don't play to the strengths of an LLM.

[Pwuts]

Why are the unit tests getting canceled?

[ntindle]

@richbeales in tests/utils.py theres a new decorator to require an API key to exist before running the test that uses it. Can you add it to the tests as required here? You can see it used in https://github.com/Significant-Gravitas/Auto-GPT/blob/master/tests/integration/goal_oriented/test_write_file.py

[github-actions]

This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request.

[github-actions]

Conflicts have been resolved! 🎉 A maintainer will review the pull request shortly.

[codecov]

Codecov Report

Patch coverage: 76.47% and project coverage change: +0.99 🎉

Comparison is base (2b5852f) 45.09% compared to head (6e4ae68) 46.08%.

❗ Current head 6e4ae68 differs from pull request most recent head 73349c2. Consider uploading reports for the commit 73349c2 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2896      +/-   ##
==========================================
+ Coverage   45.09%   46.08%   +0.99%     
==========================================
  Files          60       60              
  Lines        2856     2890      +34     
  Branches      472      476       +4     
==========================================
+ Hits         1288     1332      +44     
+ Misses       1463     1448      -15     
- Partials      105      110       +5     

Impacted Files	Coverage Δ
autogpt/prompts/prompt.py	58.33% <0.00%> (-8.34%)	⬇️
autogpt/setup.py	82.45% <92.85%> (+65.21%)	⬆️

... and 1 file with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.