Leveraging AI For Ren-C Development in 2026

A.I.
1

@hostilefork, you’ve spent years telling me to read the source. :slightly_smiling_face:

Given what Boris has done with Red/Sensei, why not do the same for Ren-C?

A small agent could index the Ren-C source, docs, issues and discussions, then expose a search/skill interface over it. The LLM can now do the source-reading that humans are so bad at keeping up with.

It doesn’t seem as though inference cost needs to be a major obstacle either — you could use something relatively cheap like Kimi and invoke it only when needed.

It feels a bit like a modern version of RebolBot in the Stack Overflow chat, except now the bot can actually read and continuously re-index the codebase rather than relying on whatever someone happened to teach it.

Is there a reason you haven’t done that for Ren-C?

2

An interesting thought from Carl about eating his own cake while others eat bread: perhaps we should be asking why the LLM is using Python via the WSL shell when it could be using Linux R3 — or Ren-C itself.

If you gave us the Ren-C semantics for the sort of things we’ve been doing — Git updates, diffs, creating and patching files, running commands, inspecting repository state, etc. — I’d simply ask ChatGPT to generate the Ren-C scripts instead.

That would be a rather better test of whether Ren-C is up to scratch as a practical development tool than talking about it in the abstract.

1 Like