search

ubiquity / ubiquibot

Putting the 'A' in 'DAO'
https://github.com/marketplace/ubiquibot
MIT License
16 stars 59 forks source link

Review upgrade #868

Open Keyrxng opened 8 months ago

Keyrxng commented 8 months ago

Resolves #791

Quality Assurance:

logs will be removed and things will be cleaned up but what's the vibes on it so far?

netlify[bot] commented 8 months ago

Deploy Preview for ubiquibot-staging ready!

Name Link
Latest commit 152f5ae744e21ae85c8e59bec7cff856688b7c94
Latest deploy log https://app.netlify.com/sites/ubiquibot-staging/deploys/6535052b22d9df00089e8f29
Deploy Preview https://deploy-preview-868--ubiquibot-staging.netlify.app
Preview on mobile
Toggle QR Code...

QR Code

Use your smartphone camera to open QR code link.

To edit notification comments on pull requests, go to your Netlify site configuration.

Keyrxng commented 8 months ago

@pavlovcik let me know which direction to take things

I think the comment suffix should be updated as now the /review command is more targeted toward the hunter as opposed to the reviewer so advising them to ensure the linked issue spec is up to date seems redundant.

I forked from the current state of the original review pr but I've had to refactor quite a bit to get things operational such as using openai function calling, adding a few more calls in the chain, refactoring the prompts.

With this new design I feel that the comment content might be changing from the reducing noise-to-signal for the reviewer to a more comprehensive format for the hunter, more insight and explanations?

As far as the reviewer is concerned if the bot hasn't approved then they move on, can't increase signal-to-noise more than that.

Keyrxng commented 8 months ago

QA:

0x4007 commented 8 months ago

After the refactor is merged I think a lot of your code will be removed because a lot of scenarios are already handled before the command handler is called.

I think that its better to leave prompts inside of the context of their functions in order to make the codebase more readable (the prompts should not be removed from their contexts, especially if they are only being used in one place.)

Regarding prompt design, I had the best experience designing my prompt within the OpenAI Sandbox. I highly recommend you experiment there. In addition, I used GPT4 on the side to help diagnose the prompt and to improve it to get the output I want.

This is cool.

Keyrxng commented 8 months ago

That's no worries, looking forward to seeing it merged and I will take things from there.

I hadn't been on the playground in a long time until just now and it has changed a lot, pretty awesome and I'll give it a bash, the assistants stuff looks class

Prompts designing prompts, promptnception lmao. When using GPT4 what mode do you use? Web browsing is default for me

0x4007 commented 8 months ago

When using GPT4 what mode do you use? Web browsing is default for me

I did not have this issue because my interface just updated now. I just looked at the custom GPT store and found https://chat.openai.com/g/g-YyyyMT9XH-chatgpt-classic a version that allegedly does not do web browsing which seems useful.

Keyrxng commented 8 months ago

my interface just updated now.

yeah I'm liking the new ui for chat and the new default just being everything is a powerhouse 😂