[FEATURE] Add visible context window usage indicator to Claude Code UI

Resolved 💬 3 comments Opened Mar 26, 2026 by tomeller Closed Apr 15, 2026

Preflight Checklist

  • [x] I have searched existing requests and this feature hasn't been requested yet
  • [x] This is a single feature request (not multiple features)

Problem Statement

There's currently no way to know how much of the context window has been consumed during a session. Users working on complex, multi-step tasks frequently hit context limits unexpectedly, causing Claude to stall, lose track of work, or produce degraded output — with zero warning beforehand.

This is especially painful for power users running long sessions with heavy tool use (many file reads, edits, agent spawns, etc.), where context fills up fast but invisibly.

  • Session planning: Users could proactively checkpoint work, commit code, and document state before context pressure causes degradation
  • Better handoffs: Knowing you're at 80% context lets you wrap up cleanly and start a fresh session instead of pushing until things break
  • Reduced frustration: Context exhaustion currently manifests as subtle quality drops, incorrect summaries, and stalled responses — all of which feel like bugs rather than a known limit

Proposed Solution

Display a context window usage indicator in the Claude Code UI — ideally as a percentage or progress bar in the status line or header. Something like:

Context: 47% used (470K / 1M tokens)

Even a simple low/medium/high indicator or color-coded dot would be a massive improvement over the current state of flying completely blind.

Alternative Solutions

_No response_

Priority

Critical - Blocking my work

Feature Category

CLI commands and flags

Use Case Example

_No response_

Additional Context

  • The desktop app (Claude Code in terminal / VS Code) has no visibility into this at all
  • The model itself can't reliably self-report context usage
  • Even an approximate indicator would be valuable — it doesn't need to be token-exact

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗