Detailed Analysis
A persistent UI rendering bug in version 4.7 of the Claude desktop application is causing the left edge of AI responses to be clipped, cutting off the first few characters of each rendered line. The issue was first reported approximately four days after the 4.7 update and affects users across multiple operational modes — specifically both Cowork and Claude Code — indicating that the defect is not isolated to any single interface component but is instead systemic to the application's core response rendering layer. Users have observed that restarting the application offers no relief, and while manually resizing the application window temporarily restores correct rendering, the fix regresses after approximately one message exchange, suggesting a dynamic layout recalculation failure triggered by incoming response content.
The scope of the bug across both Cowork and Claude Code modes is particularly notable given the architectural design of Claude's desktop application, which integrates these distinct functional environments — AI-assisted task management and scheduling under Cowork, and coding and automation tooling under Claude Code — within a single unified window navigated by top-level tabs. Because the clipping behavior reproduces consistently across these separate modes, the root cause likely resides in a shared layout or CSS rendering component introduced or modified in the 4.7 update, rather than in any mode-specific logic. The fact that window resizing produces a temporary correction points toward a viewport or flexbox recalculation issue that initializes correctly upon a resize event but fails to reapply dynamically as content updates.
The timing and nature of this regression reflect a broader challenge in shipping desktop application updates that tightly couple AI model integrations with native UI frameworks. As Anthropic has expanded the Claude desktop application to support increasingly complex workflows — including computer use, browser control via Chrome extension connectors, and high-resolution screen interaction supporting up to 2,576 pixels on the long edge — the surface area for layout-related regressions has grown substantially. Updates that introduce new capabilities or refine display scaling behavior, particularly for high-DPI environments, carry non-trivial risk of disrupting baseline text rendering, as appears to have occurred here.
From a user experience standpoint, the defect is particularly disruptive because it degrades the readability of every response across an entire session, effectively making the application unreliable for sustained productive use. Community-identified workarounds — including repeated manual window resizing, adjusting display scaling settings, and testing across monitor configurations — are palliative rather than corrective. The absence of an official acknowledgment or patch timeline from Anthropic, four days post-report, underscores a gap in rapid-response desktop software support that typically accompanies consumer-facing application development at scale. As Anthropic continues to position the Claude desktop app as a proactive personal agent platform, reliable baseline rendering will be foundational to sustaining user trust in these more ambitious agentic use cases.
Read original article →