Yuhan
@xyhstrugglerJoined Nov 2026covers Sep 18, 2025 – Nov 17, 2026
- sessions
- 346
- tokens
- 66.1B
- active days
- 115
- longest streak
- 94 days
- estimated spend
- $24,638
leaderboard ranks
earned titles
The Hands-On Pilot
1.71 mid-run corrections per session (1 or more earns it)
The Deep Diver
55% of sessions hit a context reset (30% or more earns it)
agents, by share of sessions
- Codex 100%
fun factRan 8 agents at once.
how they work
ratios of counts, not volume- Mid-run corrections
- 1.71
- Sessions with a context reset
- 55%
- Tool-call error rate
- 7%
- Started with a plan
- 11%
- Sessions at once
- 1.7×
Recap cards
16 cardsPersona
The Night Orchestrator
346 sessions in 426 days
Most agents at once
8 at once
sessions side by side, plus 1,063 subagents
Tokens processed
66.1B
tokens, about $24,638 at API list prices
Longest streak
94 days
in a row, Aug 16 to Nov 17
Longest session
125.9 hrs
of active work in one Codex session
When they prompt
Night owl
busiest hour 21:00
Manners check
3 pleases
2 thank-yous and 3 sorries in 4,305 prompts
Danger zone
9 × rm -rf
47 force pushes and 10,398 naps. Nothing caught fire.
Time with agents
937 hrs
with an agent at work, 1,473 agent-hours in all
Favourite day
Thursday
the busiest day, 909 prompts
Test runs
8,617 runs
of the tests, 62% passed
Bash hall of fame
149,561
shell commands run by agents
Where things broke
20,494
failed tool calls, 6.9% of 298,892
Words typed
26.9K
words in prompts, 0% of them questions
Slash commands
1
slash commands typed
Deep thoughts
309,310
thinking blocks, the longest 6.4 min
Published by @xyhstruggler from their own AI Insights report. Self-reported, not verified. · see the leaderboard
Find out yours.
One command in your terminal. It reads your history on your machine and makes the recap there. Nothing is uploaded.
$ npx ai-insights@latestYour report stays on your machine. Publishing a profile like this one is optional and sends only the lines you see.