Thirty-two thousand issues in thirteen months. That's what Claude Code's GitHub tracker has accumulated since launch, with roughly 2,186 new issues landing every week. An independent analysis of the repository's issue data paints a picture of a project growing faster than its maintainers can keep up with.
Half of All Closures Are Bots
Of the ~26,000 issues that have been closed, nearly half were handled by automation rather than humans. About 9,990 (37.5% of all closures) were marked as duplicates by bots, and another 3,092 (11.6%) were closed as stale after sitting without activity. That leaves roughly 51% of closures that were possibly addressed by an actual person.
Those numbers aren't inherently damning. Every popular open-source project uses bots to manage issue volume, and Claude Code's 77,748 GitHub stars put it firmly in "firehose of reports" territory. But the ratio matters because it shapes the experience of the people filing those issues. When you spend twenty minutes writing a detailed bug report and a bot closes it two weeks later, you stop filing bug reports.
The tracker currently has about 6,400 open issues, with 3,622 carrying the "bug" label.
A Bug That Refused to Die
The analysis highlights one specific bug cluster that illustrates the pattern. Starting January 8, 2026, users began reporting that custom session titles created with /rename would vanish from the /resume picker after a few conversation turns. Over the next two months, 12 separate issues were filed about variations of this problem, generating more than 70 community comments.
Community members didn't just complain. They dug into the source code, identified exact byte offsets and function names causing the problem, and posted working workarounds. One user traced the root cause on February 6 and documented it in detail.
A fix appeared on February 23 in version 2.1.53. Three hours later, version 2.1.55 shipped a regression that broke it again. As of March 14, the bug persists. Across all 12 issues and 70+ comments, the number of responses from Anthropic staff is zero.
The Pressure Valve Problem
The author of the analysis describes the tracker as "a pressure valve that lets users feel heard without anyone actually listening." That's a harsh read, but it points to a real tension in how AI companies manage developer tools.
Claude Code is not a free hobby project. It runs on a paid subscription, and Anthropic positions it as a professional developer tool. Users filing issues are, in many cases, paying customers doing free QA work. The gap between "we ship fast" and "we respond to the people finding our bugs" is noticeable.
This isn't unique to Anthropic. GitHub Copilot's issue tracker has similar patterns, and most fast-moving AI tools struggle with community management at scale. But the 49% bot-closure rate combined with zero staff engagement on a two-month-old bug cluster is the kind of data point that erodes trust over time.
For daily Claude Code users, the practical takeaway is straightforward: if you hit a bug, check existing issues before filing a new one (since duplicates account for over a third of all closures), and don't expect a fast response. The tool ships updates frequently enough that many bugs do get fixed, but the feedback loop between "reported" and "acknowledged" has a significant gap right now.