July 26, 2026
30 days later: poof
Claude Code Deletes Your Context History from Your Device After 30 Days
Your AI coding buddy now wipes old chat memories, and the internet is split between “finally” and “uh, excuse me?”
TLDR: Claude Code now deletes old local context history after 30 days unless users change the setting. Commenters are split between praising the privacy-first move and roasting it as surprise forgetfulness that could wipe useful workflow history.
Anthropic quietly confirmed that Claude Code deletes context history stored on your own device after 30 days by default, and the reaction online was immediate: one side cheered like someone had finally taken out the digital trash, while the other acted like an assistant had walked into their office and shredded a month of sticky notes. The setting lives in Claude’s data docs, where local project history sits in ~/.claude/projects/ and can be controlled with a cleanupPeriodDays option. Translation for normal humans: the app keeps some working memory on your computer, and unless you change it, old stuff gets cleaned up after a month.
But the real show was the comments. Privacy-minded users were thrilled, calling it the rare tech default that doesn’t feel creepy. Others were far less impressed, arguing that “helpful memory” vanishing after 30 days is less privacy win and more accidental amnesia. A mini-drama broke out between the “if it’s important, save it properly” crowd and the “why is an AI tool deleting my workflow behind my back?” crowd. Then came the jokes: people compared it to a goldfish, a forgetful intern, and a roommate who rage-cleans the apartment every month. Another running gag was that developers spend half their lives begging tools to forget less, while everyone else spends half their lives begging apps to track less. In classic internet fashion, the debate somehow turned a boring retention policy into a full-on identity crisis about trust, convenience, and whether software should remember your life at all.
Key Points
- •The documentation page describes Claude Code’s data usage policies, including training policy, feedback handling, telemetry, data retention, and data access.
- •Claude Code stores local data in paths such as `~/.claude/projects/`, and the documentation references a configurable retention setting called `cleanupPeriodDays`.
- •Users can control nonessential traffic, telemetry, feedback surveys, and error reporting through environment variables such as `CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC`, `DISABLE_TELEMETRY`, and `DISABLE_ERROR_REPORTING=1`.
- •A data-flow diagram shows Claude Code interacting with Anthropic services for authentication and API access, with optional telemetry and feedback flows to third-party services including Google Cloud Storage and GitHub.
- •The documentation compares encryption and default metrics behavior across Anthropic API, Amazon Bedrock, Google Cloud’s Agent Platform, and Microsoft Foundry, including the availability of Zero Data Retention for Anthropic API.