What
One of Claude Code's URL checks has changed in these ways:
- It no longer rejects URLs for being too long.
- It no longer checks that the value is non-empty text.
- Its error messages are shorter. For example, a URL containing an unencoded
@is now reported as one that "holds a raw @". The old message said it "must not hold a raw @" and told you to write it as%40.
Why
If this check rejects a URL you enter, the error message is shorter. The message no longer suggests percent-encoding @ as %40.
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubt
Which URLs go through this check is not stated, though it may be the one used for plugin URLs.