Status bar CTX indicator stuck / not incrementing during long sessions (regression)

Resolved 💬 2 comments Opened Apr 18, 2026 by Jilvert22 Closed May 26, 2026

Summary

The CTX usage indicator in the Claude Code status bar (e.g. CTX:░░░░░░░░ 6%) appears broken after a recent update. The bar does not increment during long sessions — even after heavy tool use, extended thinking (xhigh/max effort), and large input processing, the displayed value stays near its initial low value.

Impact

  • Users can no longer judge session CTX usage from the status bar
  • Critical signal for managing context in multi-agent / long-running workflows
  • Combined with xhigh/max effort occasionally producing 10+ minute stalls, users have no visual way to anticipate context limits

Environment

  • Claude Code v2.1.114
  • Model: Opus 4.7 (1M context)
  • Effort: xhigh / max
  • OS: Darwin arm64 (macOS)

Reproduction

  1. Start a long-running session with xhigh or max effort
  2. Execute many tool calls (Read, Bash, Edit) — especially large file reads
  3. Observe status bar: CTX bar stays ~6% despite heavy context accumulation

Expected (Restoration Request)

  1. CTX bar progressively fills as tokens accumulate (prior behavior)
  2. Visual indication at meaningful thresholds (50% / 30% / 15%)
  3. Usage limit / rate-limit indicators reflect actual session state

Additional Context

This regression affects multi-agent orchestration setups (cmux / tmux) where accurate per-surface CTX monitoring is critical for handover timing decisions. Restoring the previous limit display would significantly improve operational reliability.

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗