We were looking at potential ways to use GPT in Cookbook and found an interesting user pain-point in the process.
Some context: Cookbook is a web3 developer tool that aggregates smart contracts and provides metadata, documentation, security reports, and integration tools to use these smart contracts in your project.
As we've been building Cookbook, we realized that most smart contracts running in production have incomplete documentation, or lack detail.
We put together a quick MVP for docs generated GPT3.5 and the results are pretty promising. While they definitely can have issues, on the whole, they seem to do a good job filling in empty documentation, and provide an accurate description of the code.
We'd love to try:
- generated testing suites
- generated security scans
- no-code smart contract creation
Would love to get your thoughts and feedback on how we can make this a more useful feature. Looking forward to getting it running with GPT4!
Software engineers tend to follow demand, with some lagtime. Since crypto goes through these manic bull/bear cycles, demand dramatically spikes and falls. Hard to make a career change when the next bear cycle can wipe out your prospects. As crypto stabilizes, there should be a wave of immigration into web3 engineering. You can already kind of see who these engineers will be, they're often the web2 developers who just happen to be working on web3 projects (and don't know solidity)
We work with SMBs and Startups for advanced contracts, analytics and tooling. That's really how we're able to sustainably keep Cookbook free for individual developers
This is exactly it. The way to think about it is that smart contracts are effectively server-side code that is specialized to run on blockchains. They use many of the same building blocks and can often be reused. When I want to build Auth in Nodejs, I might use Passport.js, and I often use Express.js. Often I'll use mongoose.js as well. That's effectively what we're providing with cookbook, a way to find and utilize solidity packages created by the community.
Many of the contracts have their audits listed, you can read through the audits to see whether its safe to use or not. But as always, you should make sure to do your own testing before deploying to production.
The majority of web3 developers only just recently entered the space, and are looking for simpler tools. We discovered this from a long series of interviews. The goal is to build cookbook in a web2 dev accessible way, and present all of the stats and audits and information so that they are more able to make dapps.
Openzeppelin provides a very limited number of templates. Great for importing into a smart contract and extending from, but requires a fair bit of work to get there. With cookbook, people can use/deploy completed contracts from other projects for many more usecases than openzeppelin.
It's hard to say off the top of my head, but I think having stablecoin options is important, ETH, BTC (harder to do with bitcoin wallets). price should be configurable by the seller, so it's just a simple buy button for the vender. most of the POS crypto stuff I've seen is very overbloated, I think the whole thing could be simplified to an iframe button that opens metamask. This is sort of an idea I've been toying with for some time and built a prototype some time back. that's my 2 cents
I would make an embeddable payment widget for this. Provide as an iframe and include web3 dependencies. that way users can pay directly from their metamask
I think in most cases, NFTs are clearly collectible art. It's hard to say when these projects branch out and become about more, but I'd still err on the side of not securities.
Some context: Cookbook is a web3 developer tool that aggregates smart contracts and provides metadata, documentation, security reports, and integration tools to use these smart contracts in your project.
As we've been building Cookbook, we realized that most smart contracts running in production have incomplete documentation, or lack detail.
We put together a quick MVP for docs generated GPT3.5 and the results are pretty promising. While they definitely can have issues, on the whole, they seem to do a good job filling in empty documentation, and provide an accurate description of the code.
We'd love to try:
- generated testing suites
- generated security scans
- no-code smart contract creation
Would love to get your thoughts and feedback on how we can make this a more useful feature. Looking forward to getting it running with GPT4!