Hacker News Reader

Comment by gonzalohm | Hacker News Reader

Until the next time they push you back to Claude. At this point, I feel like this has to be the most unstable technology ever released. Imagine if docker had stopped working every two releases

View on HN

There is zero cost to switching ai models. Paid or open source. It's one line mostly.

What about your chat history? That has some value, at least for me. But what has even more value is stable releases.

I think this is more about which model you steer your coding harness to. You can also self-host a UI in front of multiple models, then you own the chat history.

for me there is zero value there.

You can output it as a memory using a simple prompt. You could probably re-use this prompt for any product with only slight modification. Or you could prompt the product to output an import prompt that is more tuned to its requirements.

e.g. https://claude.com/import-memory

This is one of the many reasons I don't think the model companies are going to win the application space in coding.

There's literally zero context lost for me in switching between model providers as a cursor user at work. For personal stuff I'll use an open source harness for the same reason.

I don't see any value in chat history. I delete all conversations at least weekly, it feels like baggage.

Codex doesn't read Claude.md like Claude does. It's not a "one line" change to switch.

ln -s CLAUDE.md AGENTS.md

There's your one line change.

That doesn't handle Claude.md in subdirectories. It does handle Claude.md and other various settings in .claude.

I have a CLAUDE.md symlinked to AGENTS.md

You mean Anthropic are the only ones refusing the de-facto standard despite a long-standing issue: https://github.com/anthropics/claude-code/issues/6235

And as others have said, it's a one-line fix. "Skills" etc. are another `ln -s`

Reply on news.ycombinator.com