scastillo

It seems to me that skills are the same as projects on the web interface, but with executable files

It's better if you think that the only thing that is really there is a context window.

They can add complex concepts and tools on top, but all that is is a different way to put things in the context window. Even the chat history on the web... You are not sending a message every time... It's not really a chat... the model is writing what it predicts will come next, like autocompleting a Word document that is written in a chat-like format.

So agents are like you, opening a new window and having the chat there, so you don't pollute the current window with all the tokens needed to process that question, and to use only the output here.

This is important bc of the effective context window problem. Models are more accurate the smaller the context is.

Hence, MCP tools are problematic. If you have registered many of them, the rules for using each one are added to your context, even if you don't use them.

Having a very extensive Claude.md file is also problematic.

You can use skills to instruct the model on which agents to use when requesting a specific thing. Antrophic says they have trained the model to discover on its own when to read the skill and follow the instructions you picture there, which can include Python scripts to run.

So yeah, agents help the model save context window for your current problem, skills help the model follow your instructions better, and instructions can include agent calling, and MCP is crap, you'd better ask the model to generate code to make that call

Oh, there are also slash commands. I don't really use them... if someone has a success story for them, I would love to know about it.

father time? time by itself does not explain anything... is the new evidence that happens over time the one that teaches us. We are just doomed to a mostly monotonic growing time function. So we need action not that invitation to sit down and wait for the next time chunk to hit.

Usually this means you may probably get distracted easily. So you are good at maths and physics, but then there is this new concept on software engineering and what you do? maybe you switch and forgot for a while about the last topic you where becoming an expert.

There is one thing you are doing constantly and that is: Learning. You are becoming one of the best at learning a new topic quickly.

So is not really as in susans blog post that we are never going to have a single destiny... our destiny is to be the best at learning new topics in record time. This is certainly useful at a research team and new knowledge areas can really laverage someone like that. It's like the category theory expert who can found relations between mathematical objects.

My mindset towards this comes from an stoic maybe buddhist kind of perspective mainly influenced by Thich Nhan Nhat: So you want to be an expert in many things but there is no time for that? cool, embrace it, hug that feeling, accept it you don't need to suffer about it and maybe you already know this and you are looking for practical advice to get more done right?

So that is what I do, I embrace this fact, breath it out, and continue with the topic at hand.

Now getting back at the distractibility-factor... the real problem is when this starts making us not deliver important things here i'm with the person who commented about systems. Atomic Habits is a good book to make those systems and make it simpler for our biased existence to follow a predesigned path daily just as if you were your own parent and designed systems for the little you, the one that have to make decisions daily.

Honestly some psilocybin have helped me understanding this... and micro-doing it seems to add clarity and helps on the letting go.

This seems to be the opposite advice from what pg is talking about. The very last paragraph closes saying don't worry about the todo just follow your muse.

If you are forcing yourself that much to get it done maybe it's not the right thing to be working on anyway?

Thanks. This is sooo pleasurable. Now add a wolfram and a GPT-3 plugin

In my experience you can't get Colombian top developers for less than 60k

Report to IRS is not a bug is a feature you have to pay taxes somewhere and there are plenty of non double taxation agreements within the US and Latam.

decrypted file is not ogg

It outready outperformed the topic of this room for example

Which may be good. If I post to HN I'm looking for critics, not for an empty Attaboy.