claude-hud

mirror of https://github.com/jarrodwatts/claude-hud.git synced 2026-05-21 07:22:44 +00:00

Files

Paul e8d64924bc fix: resilient usage display under API rate limiting (#193 )

* fix: resilient usage display under API rate limiting

The Anthropic usage API rate-limits to ~1 call per 5 minutes. With the
previous 60s cache TTL, 4 out of 5 API calls returned 429, causing the
HUD to permanently display "(429)" instead of actual usage data.

Three-layer fix:
- Increase cache TTL from 60s to 5 minutes to match rate limit window
- Preserve lastGoodData in cache across rate-limited periods so the HUD
  always shows the best available data instead of errors
- Exponential backoff (60s→120s→240s→5min cap) with Retry-After header
  support for consecutive 429 responses

Also show "syncing..." instead of raw HTTP status on first-run rate limit.

* Update usage-api.ts

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

* fix: harden 429 cache fallback behavior

* test: stabilize usage cache suite after rebase

---------

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Jarrod Watts <jarrod@cubelabs.xyz>

2026-03-14 11:11:36 +11:00

fixtures

fix(context): scale autocompact buffer by raw usage (#190 )

2026-03-14 10:57:49 +11:00

config.test.js

Add configurable usage cache TTL settings (#175 )

2026-03-09 15:13:22 +11:00

core.test.js

fix(context): scale autocompact buffer by raw usage (#190 )

2026-03-14 10:57:49 +11:00

extra-cmd.test.js

fix(extra-cmd): improve robustness and add tests (#94 )