Repository navigation
Context Monitor
Real-time visibility into your Claude Code session directly in VS Code — no extra API calls, no rate limit risk.
- Token breakdown — Input, Output, Cache read, Cache write with visual bars
- Estimated cost — Per-model pricing (Opus, Sonnet, Haiku) calculated automatically
- Session stats — Response count, session duration, total tokens
- Models used — Badge showing which models were used in the session
- Multi-session support — Dropdown selector when multiple sessions are active in the same workspace
- Status bar — New item showing session cost and tokens at a glance: $39.52 · 19.7M tok
- Auto-refresh — Watches the active transcript file and updates automatically
- Cross-platform — macOS, Linux, and Windows
All data is local (reads Claude Code's JSONL transcripts) — zero extra API calls.
Rate Limit Protection
- Automatic retry with exponential backoff (1s/2s/4s) on HTTP 429 and 5xx
- Respects retry-after header from Anthropic's API
UI Improvements
- Simplified usage bars (session 5h + weekly 7d) like the Anthropic web console
- Context metrics visible even without API authentication
- Auth guidance: hint to run claude auth login
Breaking Changes
- ludevMetrics.barStyle setting removed (consolidated to single style)
- ludevMetrics.viewMode setting removed (Extended/Compact toggle)
New Settings
| Setting | Default | Description |
|---|---|---|
ludevMetrics.costRates.inputPerMtok |
3.00 | Cost per million input tokens (USD) |
ludevMetrics.costRates.outputPerMtok |
15.00 | Cost per million output tokens (USD) |
ludevMetrics.costRates.cacheReadPerMtok |
0.30 | Cost per million cache-read tokens (USD) |
ludevMetrics.costRates.cacheCreationPerMtok |
3.75 | Cost per million cache-creation tokens (USD) |
Removed Settings
| Setting | Reason |
|---|---|
ludevMetrics.barStyle |
Consolidated to single style |
ludevMetrics.viewMode |
No longer needed |