If you have any luck, let me know.

Part of the reason why I wanted to try and let Claude have direct access to
VirtualT, it because it was really struggling to debug things when the ML
subroutines went wrong.

It had a tendency to want to try and cache pointers, and by the time it got
around to using them, they had drifted. Instead of jumping into its
subroutine, it would just jump into some random garbage.

So far, I’ve wasted a lot of time and tokens on reminding it to not make
the same mistakes over and over.

Today, I’ve wasted most of my tokens on getting it to navigate VirtualT,
upload a .DO, enter BASIC, LOAD”prog.do” to tokenize, and RUN.

The problem is, while Claude Code and Codex are perfectly fine at using the
MCP as a part of a single pre-defined plan, it’s not really good at
continuing without the human.

So the entire testing plan needs to be defined up front. It makes it less
agentic and more human-in-the-loop than I would have liked

-George

On Wed, Dec 24, 2025 at 1:37 PM John Hogerhuis <[email protected]> wrote:

> On Wed, Dec 24, 2025, 9:18 AM [email protected] <[email protected]>
> wrote:
>
>> Hi John,
>>
>> This is basically (no pun intended) what the github repo that I linked,
>> represents.
>>
>> A SKILL.md is meant to be all of the relevant context an Agent needs to
>> perform a task, as well as any tooling(scripts) and usage examples.
>>
>> In there you will find markdown which represents pretty much all of the
>> above you listed.
>>
>
> Yes. And there is a lot of stuff one could add.
>
> Here's an article with some efficiency tips.
>
> I guess each vibe coder can tailor their rules and get a edge on the
> competition ;-)
>
> https://bitchin100.com/wiki/index.php?title=BASIC_Efficiency_Tips
>
> Also it would be good to have content to teach it to efficiently directly
> executive ML code in basic as strings and REMs. And PIC and relocatable
> routines. I have some articles on that.
>
> -- John.
>
>>

Reply via email to