Since you seem to be mentioning 'destiny', 'life arc' and 'advice' -it may be wise to put in some kind of a legal disclaimer for your own sake -- the use of LLM toys for advice about life and similar have led to some really bad outcomes, even suicides as we know now.
I once did an exercise to review the previous week's horoscope and grade each sign on how closely it corresponded to my actual week. I found no consistency whatsoever, so I'm not just miscategorized into the wrong sign, but in fact none of them had any predictive value. It might be fun to broaden that out to other sources of horoscope than my local Coffee News, maybe a better mystic is what I need
So the astrology is deterministic, and the interpretation is probabilistic. Which is interesting, because that’s exactly where most of the perceived “accuracy” tends to come from.
This is for Claude Code. It's not hard to install/run Claude Code but it might be a little intimidating if you're not used to command lines.
I think in the regular Claude app, you can set up something like this as a "project" with a custom prompt. Claude might be able to pull the prompting from Github and convert it.
I actually did not use prompts here. It is actual calculation from the python lunar and it cast the result to LLM which you use. The LLM has the old books such as giving you your fortune following lunar year calculation. Have a fun with it!
I would recommend you consider a Claude code hook, you could then specify a particular part of the agent lifecycle to invoke the fortune as an alternative to manual invocation of the skill.
You could do it on `SessionStart` to give a horoscope.
I am looking at making a skill to test random programming knowledge, so I don't forget that an Integer is 32bits (at least in Java)
People have been goaded into killing themselves or others by LLMs so I don’t take kindly to third rate hucksters abusing the behaviours that lead to it with these tools for a laugh.
Have you ever had to deal with someone whose brain has been completely cooked by LLM induced psychosis? Because I have — it destroys people’s lives completely unnecessarily. So I will not apologize for having a visceral disgust towards people and behaviours that will amplify these very real harms for their own personal gratification and online clout chasing.
Since you seem to be mentioning 'destiny', 'life arc' and 'advice' -it may be wise to put in some kind of a legal disclaimer for your own sake -- the use of LLM toys for advice about life and similar have led to some really bad outcomes, even suicides as we know now.
If you act like a doctor and someone gets hurt from your acting, that's the main trouble.
I am not sure the same standard would apply for a fortune teller :)
fair points. I will take a look at it
I am resisting the urge to be very snarky.
But how do you know that Claude's hallucinated horoscopes are any less accurate than traditional techniques? Have you A/B tested them?
I once did an exercise to review the previous week's horoscope and grade each sign on how closely it corresponded to my actual week. I found no consistency whatsoever, so I'm not just miscategorized into the wrong sign, but in fact none of them had any predictive value. It might be fun to broaden that out to other sources of horoscope than my local Coffee News, maybe a better mystic is what I need
[flagged]
[dead]
So the astrology is deterministic, and the interpretation is probabilistic. Which is interesting, because that’s exactly where most of the perceived “accuracy” tends to come from.
Could you flesh out the installation instructions a little more? I haven't used the plugin marketplace yet, and don't have a /plugin command.
Claude itself is not able to tell me how to install it.
This is for Claude Code. It's not hard to install/run Claude Code but it might be a little intimidating if you're not used to command lines.
I think in the regular Claude app, you can set up something like this as a "project" with a custom prompt. Claude might be able to pull the prompting from Github and convert it.
There is a claude code in the claude gui as well, i think everyone must be saying to use the tui.
[flagged]
Just edited readme as well
[dead]
Curious what prompt boundaries you're using. It's the kind of thing that's fun until it quietly invents details.
I actually did not use prompts here. It is actual calculation from the python lunar and it cast the result to LLM which you use. The LLM has the old books such as giving you your fortune following lunar year calculation. Have a fun with it!
Looks like a supply chain risk to me.
I did not understood the meaning of your words.
I thought anthropic is tightening usage but it seems people have nothing to do with their tokens …
I'm doing what people wants. Thanks!
I would recommend you consider a Claude code hook, you could then specify a particular part of the agent lifecycle to invoke the fortune as an alternative to manual invocation of the skill.
You could do it on `SessionStart` to give a horoscope.
I am looking at making a skill to test random programming knowledge, so I don't forget that an Integer is 32bits (at least in Java)
Auto-run on session start (optional) added. Please check it out!
Yes this is exactly what we need — the hallucinating dunning-Krueger inducing psychosis machine spewing made up pseudoscientific nonsense. Bravo.
haha great prose
People have been goaded into killing themselves or others by LLMs so I don’t take kindly to third rate hucksters abusing the behaviours that lead to it with these tools for a laugh.
I don't think it's that deep.
Seems like a bit of an overexageration to a bit of fun.
Have you ever had to deal with someone whose brain has been completely cooked by LLM induced psychosis? Because I have — it destroys people’s lives completely unnecessarily. So I will not apologize for having a visceral disgust towards people and behaviours that will amplify these very real harms for their own personal gratification and online clout chasing.
I don't think its appropriate for you to lash out at others because of your own life issues.
I don’t care.
He is 100% ironically using an LLM to comment.
Nope, 100% character by character ranting and raving.
Hope you enjoy your life and time.
Isn't that the memory files are loading by default?
interesting, but I really wonder how big the audience is of people wanting to know their fortune while claude coding within the claude code interface
This is my fun project! Not a profitable entity. Thanks
for what it's worth I think this is fun. I'm not into this sort of thing but it's fun none the less.
This was what I was targeting. Thanks
Love to see the evals on this one!
i wonder how the output will be in different LLM
feature request: marriage palace built using two user profiles
/destiny compat <partner birth> <city> <m|f>
do it now.
Now the call center cranks will have even better scripts to read. Perfect...
Let me have fun with it. Thanks
I'm not trying to spoil it!
[flagged]
[dead]