v5.3.0

What's Changed

Improvements

• Live effort detection via stdin effort.level (CC 2.1.119+) — the canonical source. The model badge now reflects mid-session /effort changes and correctly displays all levels including max.
• Effort badge is now omitted entirely on models that don't support effort, instead of showing a fallback character.

Display Changes

• model:effort=short now renders L/M/H/X/MAX (was L/M/H/A with first-letter fallback for unknown values).
  • xhigh → X (explicit, was H via fallback)
  • max → MAX (was lowercase max)
  • auto no longer appears — CC always resolves it to a concrete level
• model:effort=full now renders the actual level name including xhigh / max (was low/medium/high/auto).

Removed

• CLAUDE_CODE_EFFORT_LEVEL env var override and settings-file fallback. These were workarounds for a missing upstream feature. No user action required — stdin now provides the live level automatically.

Full Changelog: v5.2.1...v5.3.0

v5.2.1

What's Changed

Fixes

• Python 3.10 compatibility: replaced datetime.UTC (3.11+) with datetime.timezone.utc (available since 3.2). The script now runs on Python 3.10 — e.g. the default interpreter on Ubuntu 22.04 LTS. README updated to note the tomli package fallback for custom themes on 3.10.

Thanks to @vzoha for the contribution (#3)!

Full Changelog: v5.2.0...v5.2.1

v5.2.0

What's Changed

New Features

• Plugin system for custom segments: Extend the statusline with custom segments by dropping .py files into plugin directories (.claude/statusline/ project-level, ~/.claude/statusline/ global). Each plugin defines register(api) to register segments via api.add_segment(). The API provides fg(), bg(), text_color(), RESET, and BOLD for themed ANSI output. Plugin errors are silently ignored. See README for a full example.

Improvements

• Raw JSON input from Claude Code now available to segment renderers via ctx["data"] (enables access to session_id, cwd, and other fields)
• Code style cleanup and formatting improvements

Full Changelog: v5.1.0...v5.2.0

v5.1.0

What's Changed

New Features

• Effort level display in model badge via model:effort=short (L/M/H/A) or model:effort=full (low/medium/high/auto). Reads from settings files with precedence chain. Note: max is session-only and cannot be detected.
• Dump mode (SL_DUMP=1) for development: logs every stdin JSON input with timestamp to /tmp/claude-statusline-dump.jsonl
• Native rate_limits support: reads usage data directly from CC 2.1.80+ stdin JSON, eliminating OAuth API calls. Falls back to deprecated OAuth API for older CC versions.

Full Changelog: v5.0.0...v5.1.0

v5.0.0

What's Changed

⚠ Breaking Changes

model:effort option removed

• Claude Code now displays reasoning effort level natively
• If your SL_SEGMENTS includes model:effort=short or model:effort=full, remove the :effort=... part

Removed

• get_effort_level() function and CLAUDE_CODE_EFFORT_LEVEL env var reading — no longer needed

Full Changelog: v4.13.1...v5.0.0

v4.13.1

What's Changed

Fixed

• Burndown noise suppression after weekly window reset — the relevance filter now scales inversely with the Bayesian trust factor, preventing misleading "may run out X sooner" warnings from stale or spiked utilization data in the first hours of a new window

Full Changelog: v4.13.0...v4.13.1

v4.13.0

What's Changed

New Features

• Worktree segment: Displays worktree info in {curly braces} when running in a --worktree session. Only renders when worktree data is present — no output in normal sessions.
  • Default shows worktree name: {feature-auth}
  • Configure with worktree:show=name,branch for multiple fields
  • Supports name, branch, path, origin (with ~ home shortening)
  • New text_worktree theme color (Nord15 purple, customizable)
  • Included in default segments after directory

Full Changelog: v4.12.0...v4.13.0

v4.12.0

What's Changed

Removed

• Legacy context window fallback code — transcript parsing, context_na_message segment, build_na_line(), and fallback option for percentage/tokens segments have been removed. Claude Code now provides used_percentage reliably, making these mechanisms unnecessary. If used_percentage is missing, the status line produces no output instead of showing an N/A fallback.

Full Changelog: v4.11.0...v4.12.0

v4.11.0

What's Changed

New Features

• Reasoning effort level in model badge: Shows the current reasoning effort (high/medium/low) next to the model name
  • model:effort=full (default) — full word: Opus 4.6 high
  • model:effort=short — single letter: Opus 4.6 H
  • model:effort= — hidden (previous behavior)
  • Reads from CLAUDE_CODE_EFFORT_LEVEL env var, ~/.claude/settings.json effortLevel, or defaults to high

Full Changelog: v4.10.0...v4.11.0

v4.10.0

What's Changed

New Features

• Added directories segment: New added_dirs segment shows directories added via Claude Code's /add-dir command
  • Sorted alphabetically, displayed in muted gray
  • Default separator: • (configurable via separator option)
  • Supports basename_only=1 to show just directory names
  • Included in default segments after directory

Full Changelog: v4.9.0...v4.10.0