I would like to pour some proverbial gasoline on the Crystal language and it’s development team/contributors.
For the month of February, I a willing to sponsor/reimburse up to 3 people who are willing to use the extra power for targeting and improving the Crystal language with some of the bigger wish list items that would help bring Crystal up to par or even exceed other languages.
I think that with the right team members being empowered and unlocked by a tool and model as powerful as Opus 4.5 and Claude Code as a harness, we should see a greater level of development than was possible.
I want to be clear, I am willing to sponsor the $100/mo Max subscription so that you can get meaningfully higher usage limits. I currently have the $200/mo plan myself and I’ve never been able to use 50% of my overall limit.
I am also willing and eager to teach others how to setup your coding assistant so that it has memory to learn and memorize it’s mistakes, along with some other features that have been general development and working easier.
The kinds of things I’d love to see are:
-
An improvement to shards. I’d like to see the shards package management updated so that we can use a post-install script either on purpose or opt-in. I know that it’s been discussed. I know that we have plenty of reasons for disliking the always run post install script. So, somebody who could tackle this and make it meaningfully better so that we still have this capability, but the overall experience flows better, is documented, and ideally can be used with an AI model to ensure that local coding assistants at least understand what the tool is capable of.
-
Working on RFC ideas, such as support for web assembly, including the web assembly with garbage collection. Have basic support for WASM now, but if we could take an AI and work through the entire RFC in order to add a full WebAssembly target to Pile 2, I think that we would be unlocking a very significant next step for the crystal language.
-
Improving the compiler by investigating hard tasks like partial recompilation. We know it’s possible, but requires a delicate and detailed hand to know when the model is going off the rails with it’s ideas or not. This is a good use case right now for what the models can accomplish right now.
-
After doing these library-level improvements kind of tasks, it would be really nice to also see that first layer of shards maybe receive some love that they need. So I’m thinking of like the MySQL gem. Maybe we finally actually commit to adding support for the modern authentication for a MySQL MySQL compatible RDBMS.
-
It would be really nice to see the documentation updated, even if it’s something where somebody goes through and performs an audit to get a much better idea of all of the gaps we may have. I think a good way to do this would be to have Claude just read through all of the issues and do some categorizations and then look for things that people ask about that aren’t addressed on the docs, but are a language feature. This is the kind of thing that Claude can actually do relatively well, especially with the Ralph Wiggum loop. And then you could build that database probably in a couple of hours or less and still have enough to then come up with a process to actually maintain this new information super highway.
-
It would be absolutely epic if somebody could research and organize a way to include coding assistant files for your library when someone installs it with shards. So let’s say I write an AI development shard and I need to, or I would like to, provide documentation for that to anyone who installs it. I don’t necessarily want to have to run an MCP for everyone’s coding assistants to access to read the docs. It is possible to just let people crawl the website and have their agents ingest it from there. But it would also be really nice to just provide a static fixed version that people can have access to while offline. It would also be much cheaper than trying to run an MCP ourselves for every thing. Even a consolidated MCP server would become a pain to maintain over time and not something I would want to do for free for now.
Are there a couple people who want to take on these really big hard tasks?
I’m going to be clear, I prefer to sponsor anyone who’s already been contributing to the language on a more regular basis. I also want to plan this so that a group of us can collaborate together because I know that I have insights on how coding assistants work that make me significantly better as a developer and the quality is maintained at the level that I am happy with. I think if I’m able to share this with others, give them the quality of the tool that I’m working with, which is Claude Code Max, then we can see these other already incredibly intelligent people excellent exponentially grow themselves.