Live data from Hacker News

Learn Claude Code by doing, not reading

claude.nagdy.me

101–110 of 133 posts

Re: Learn Claude Code by doing, not reading

#101

Earlier quoted context omitted.

I noticed 1M context window is default and no way not to use it. If your context is at 500-900k tokens every prompt, you’re gonna hit limits fast.

do you pay for the full context every prompt? what happened with the idea of caching the context server side?

You don't. Most of the time (after the first prompt following a compaction or context clear) the context prefix is cached, and you pay something like 10% of the cost for cached tokens. But your total cost is still roughly the area under a line with positive slope. So increases quadratically with context length.

Re: Learn Claude Code by doing, not reading

#102
post #52

Earlier quoted context omitted.

I'd like to use Opus with OpenCode right now to combine the best TUI agent app with the best LLM. But my understanding is Anthropic will nuke me from orbit if I try that.

OpenCode with a Copilot Business sub and Opus 4.6 as the model works well

I'm looking at their plans (https://github.com/features/copilot/plans) it seems like the limits might be pretty low, even with the Pro+ plan which is 2x the cost of Claude Pro. It seems like Claude Pro might be 10-20x the Opus tokens for only twice the price.

Re: Learn Claude Code by doing, not reading

#103
post #12

Side note: I don’t know what Anthropic changed but now Claude Code consumes the quota incredibly fast. I have the Max5 plan, and it just consumed about 10% of the session quota in 10 minutes on a single prompt. For $100/month, I have higher expectations.

[flagged]

Re: Learn Claude Code by doing, not reading

#104
post #12

Side note: I don’t know what Anthropic changed but now Claude Code consumes the quota incredibly fast. I have the Max5 plan, and it just consumed about 10% of the session quota in 10 minutes on a single prompt. For $100/month, I have higher expectations.

Have had similar issues with costs sometimes being all over the map. I suspect that the major providers will figure this out as it’s an important consideration in the enterprise setting

Re: Learn Claude Code by doing, not reading

#105
post #7

Strongly agree with the sentiment, but I'd say if you're familiar with the terminal you may as well just install it and truly 'learn by doing'! I could see this being great for true beginners, but for them it might be nice to have even some more basics to start (how do I open the terminal, what is a command, etc).

[flagged]

Re: Learn Claude Code by doing, not reading

#106
post #60

Earlier quoted context omitted.

I'd like to use Opus with OpenCode right now to combine the best TUI agent app with the best LLM. But my understanding is Anthropic will nuke me from orbit if I try that.

You can use Opus with OpenCode anytime you want, just not with the Claude plan. You can use it via API with any provider, including Anthropic's API. You can use it with Github Copilot's plan. The only thing you can't do without getting banned is use OpenCode with one of Claude's plans.

I keep seeing this "you can use the inconvenient and unpredictably costly way all you want" pedantic kneejerk response so often lately.

It's like saying well humans can fly with a paraglider. It is correct and useless. Most here won't have cash to burn with unbounded opus api usage.

Re: Learn Claude Code by doing, not reading

#107

I use claude code every day, I've written plugins and skills, use MCP servers, subagent workflows, and filled out the "Find your level" quiz as such. According to the quiz, I am a beginner!

I think it’s just buggy, I had the same results despite of knowing every single question in depth other than building a plugin.

Re: Learn Claude Code by doing, not reading

#109
No. 100% no. Learn the art of programming. Read K&R. In 5 years we will see "new is old" again. Tokens will become prohibitively expensive and, once more, another $steve.ballmer.2.0 will be yelling "developers ... developers". And Claude Code ... will become another "pentesting" / "linting" tool.
Post reply on HN