"Why this is more than an inconvenience. For paying users who treat their sessions as intellectual property — design reasoning, prompt history, hard-won context — this is silent, unconsented destruction of user-owned data. The transcripts are the user's record of their own thinking and work; deleting them by default, silently, with no recovery, inverts the expected ownership relationship. A 30-day default that quietly discards months of accumulated reasoning is a poor default for that audience, however reasonable it is for disk hygiene in the general case."
Those users would be wise to back those files up if they consider them valuable intellectual property. If they're important enough that you'd miss them after a disk failure, then they should have been being backed up already.
>if they consider them valuable intellectual property [...]
It's hard to take any of what's written seriously, given that it's all AI generated. Did the user actually lose "valuable intellectual property", or did they tell claude to write as dire of a justification as possible?
Isn’t “Anthropic won’t fix it” a little sensational for barely month old issue with little activity(two upvotes, one of which is me), in backlog of 5k+? Agree that it’s a real issue that need fixing however
- it is silent, not announced anywhere, and calls rm on your data
- setting the days to zero doesn't disable it, it immediately rm's ALL your chats
- even setting it to _something_ big doesn't prevent data loss if you launch Claude Code in some modalities (certain subagents, etc) that don't load this config key but perform deletion anyway (with the 30 day default)
I just noticed that it’s a config option. Weird that it’s so short though, I can understand why it may be needed for users who spawn hundreds of sessions a day
Wow good timing, I’ve been working on a session hub of sorts for devs. Check it out, you can store your sessions on here. https://joe-store-frontend.onrender.com
This is clearly shown in the settings/config (or at least was last time I looked), if people are surprised by this I recommend asking claude code what settings you can tweak
The best mitigation I've found against this is training Claude to collate what it does within the project dir, specifically a CLAUDE.md vision file and .claude/changelog that documents the changes it makes. The biggest pain point though is remembering to force it to do that between sessions (man is that contextual memory unreliable sometimes).
Oh yeah - it commonly doesn't update my BACKLOG.md after shipping something. It tends to catch itself on the next set of work, but sometimes I have it clean up based on recent commits.
Except that those customers can access the traces for 30 days, and freely copy them at any point during that period? It's a usability issue at worst, not some anti-customer conspiracy.
Those users would be wise to back those files up if they consider them valuable intellectual property. If they're important enough that you'd miss them after a disk failure, then they should have been being backed up already.
It's hard to take any of what's written seriously, given that it's all AI generated. Did the user actually lose "valuable intellectual property", or did they tell claude to write as dire of a justification as possible?
https://github.com/anthropics/claude-code/issues/4172
or this one from December, still open:
https://github.com/anthropics/claude-code/issues/9258
Some more fun facts about this:
- it is silent, not announced anywhere, and calls rm on your data
- setting the days to zero doesn't disable it, it immediately rm's ALL your chats
- even setting it to _something_ big doesn't prevent data loss if you launch Claude Code in some modalities (certain subagents, etc) that don't load this config key but perform deletion anyway (with the 30 day default)
Ccrider is the one I use: https://github.com/neilberkman/ccrider
Session backups live outside of ~.claude so are not deleted and can be resumed.
Imagine taking a leave, come back in 30 days, boom. Your transcripts are rm'd, silently, without any notification or confirmations.
Bonus: setting it to 0 days rms everything immediately
I think it's more like, "We don't want our harness to need to be able to interpret every version of our conversation format indefinitely."
which makes anthropic on the whole a lot less trustworthy to have access to any machine
Film at 11.