Real coding sessions · never self-reported
Where the tokens go
50% building · 949.0M classified tokens
25 builds predate classification and aren't included.
Fixed the weather world's puddle reflections under storm lighting
1.0M tokens vs typical 185.3K
Fixed puddle reflections to display sky color instead of black during dim lighting scenes
Fixed tab label alignment in the favorites feature
14.2M tokens vs typical 185.0K
Fixed tab labels that were misaligning in favorites.
Fixed weather brief calculations for storms and wind
6.1M tokens vs typical 185.0K
Fixed storm and wind timing calculations, verified all fallback paths work correctly.
Debugged meteo-brief
1.4M tokens vs typical 183.7K
6h 22m session running commands and testing. 6.1M tokens across 14 prompts using claude-sonnet-5.
Debugged meteo-brief
8.5M tokens vs typical 183.7K
6h 22m session running commands and testing. 6.1M tokens across 14 prompts using claude-sonnet-5.
Autonomous session on que-fer
4.0M tokens vs typical 183.7K
Autonomous 14s agent session using claude-opus-5. 0 tokens processed.
Refactored profile page with consolidated tabs
3.1M tokens vs typical 179.5K
Extracted shared tab logic, added unit tests, and verified both profile routes render correctly.
Autonomous session on que-fer
2143 sessions tracked this week across all builders
Autonomous 17s agent session using claude-opus-5. 0 tokens processed.
Autonomous session on que-fer
870 sessions tracked this week across all builders
Autonomous 17s agent session using claude-opus-5. 0 tokens processed.
Autonomous session on que-fer
11.6M tokens vs typical 179.5K
Autonomous 17s agent session using claude-opus-5. 0 tokens processed.
Autonomous session on que-fer
9.2M tokens vs typical 179.5K
Autonomous 17s agent session using claude-opus-5. 0 tokens processed.
Designed profile and favorites architecture; corrected caching assumptions
1.2M tokens vs typical 179.5K
Corrected caching assumptions and verified the favorites architecture approach.