Claude Code users face unexpected data loss as transcripts are erased every 30 days without clear notification. A lack of transparency from Anthropic raises security and usability concerns.
Claude Code users are discovering an unexpected and frustrating issue, conversation transcripts are silently disappearing after a 30-day period. This automatic deletion feature, meant as a security measure, isn't clearly disclosed to users, leading to confusion and potential data loss.
The Hidden Configuration #
The core of the issue lies in the cleanupPeriodDays configuration setting, which defaults to 30 days. Each time the app starts, it erases any.jsonl files that exceed this age limit. Remarkably, this information is buried in documentation, with no direct warning provided to users. Anthropic, the company behind Claude Code, maintains that this policy has been in place since launch, aiming to protect against security and privacy risks associated with storing transcripts indefinitely.
While security is a valid concern, the lack of transparency is troubling. Users aren't given any install-time alert or initial run dialog that this data clearing practice exists. This oversight leaves many unaware until it's too late, resulting in lost coding sessions and the associated context that's vital for ongoing projects.
Users Demand Better Communication #
On GitHub, users have expressed their dissatisfaction. One user, FTSBrand, highlighted the absence of any initial disclosure about the cleanup feature, calling it a misunderstanding between what users perceive as durable working knowledge and the reality. Others have noted the loss of valuable context, such as design discussions and debugging analysis, which are irreplaceable artifacts in research and development work.
The expectation that users should back up their data independently is a weak defense for what's essentially a product oversight. Good backups are indeed prudent, but they don't compensate for the lack of upfront communication regarding such a significant data retention policy. As GitHub user caioribeiroclw-pixel pointed out, clearer product-level disclosure is necessary.
The Technical Hurdles #
Attempts to alter the retention period haven't proven successful for all. The deletion process is tied to a transcript's modification time rather than the last activity timestamp. This means that any script or client that interacts with file timestamps could inadvertently trigger the deletion of recent sessions, as explained by user ojura.
This highlights a broader issue: the need for software developers to prioritize transparency and user control over data retention settings. Why should users face the risk of unintentional data loss due to insufficient product communication? The stakes are high, particularly when dealing with sensitive information or critical project histories.
The need for a fix is evident. While the security rationale is understandable, the lack of proactive communication and user-friendly options for data management isn't. The reserve composition matters more than the peg, and in this case, clarity in data retention policy could significantly enhance user trust and application usability.
Get AI news in your inbox
Daily digest of what matters in AI.