@kevalprajapati
#8this weekJoined Jun 14, 2026
186.0MtokensCaptured live
291builds
8660.5hbuild time
138.1klines
19dactive 30d
12wweek streak
Real coding sessions · never self-reported
Highlights
Where the tokens go
54% building · 186.1M classified tokens
Building100.8M54%
Delegating36.6M20%
Thinking26.7M14%
Verifying22.0M12%
31 builds predate classification and aren't included.
12
week streak
25.2M
token record
6d ago
214
prompt record
7d ago
5d ago
Debugged aai
aai
First session with opus-5[1m]
First build using this model
451.2ktokens
7prompts
1:22:41time
257lines
Claude Codeclaude-opus-5[1m]Python$9.00
balanced mix
1h 22m session running commands and testing. 451K tokens across 7 prompts using claude-opus-5[1m].
6d ago
Shipped demo seeder with realistic applicant scenarios
kore
MARATHON
Biggest session this week
Previous best: 3.8M · +369%
25.2Mtokens
32prompts
46:24:29time
1,672lines
Claude Codeclaude-fable-5TypeScript$1322.71
mostly delegating
Shipped a demo data seeder to quickly test the full end-to-end workflow with realistic applicants
6d ago
Verified bilateral tongue model status and integration
aai
ONE-SHOT
Top 3% build time this week
1953 sessions tracked this week across all builders
113.8ktokens
1prompts
27:00:16time
Claude Codeclaude-opus-4-8Markdown$1.42
mostly verifying
Determined renderer integration status of bilateral model channels and documented findings.
6d ago
Built pricing and comparison framework for models
nucleus
LOCKED IN
First session on nucleus
First build tracked for this project
512.8ktokens
19prompts
40:22:19time
1,604lines
Claude Codeclaude-opus-4-8TypeScript$11.45
mostly building
Implemented pricing catalog, fixed model ID mappings, and built evaluation framework.
6d ago
Built tailored and ATS-verified resume for job application
resume system
Top 2% build time this week
1949 sessions tracked this week across all builders
463.7ktokens
4prompts
27:00:16time
307lines
Claude Codeclaude-fable-5Markdown$17.71
mostly building
Generated tailored resume with full ATS verification for job submission.
7d ago
Debugged and halted a runaway network monitoring loop
keval
Most prompts this week
214 prompts
427.9ktokens
214prompts
37:01:02time
229lines
Claude Codeclaude-opus-4-8Shell$9.29
mostly building
Diagnosed a runaway recurring job and cancelled it to halt the cascading API errors.
7d ago
Set up project foundation with spec and developer guide
wedding-invite
Top 1% build time this week
1944 sessions tracked this week across all builders
189.0ktokens
3prompts
65:00:11time
450lines
Claude Codeclaude-opus-4-8Markdown$2.06
mostly verifying
Initialized project workspace with specification, getting-started guide, and development conventions.
7d ago
Built invitation templates and editor with live preview
wedding-invite
DEEP WORK
Biggest session this week
Previous best: 1.2M · +219%
3.8Mtokens
15prompts
65:00:14time
12,068lines
Claude Codeclaude-opus-4-8TypeScript$195.26
mostly building
Implemented templates system, invitation editor with live preview and autosave, and account dashboard.
8d ago
Added 5 sample templates with real preview images
wedding-invite
FULL SEND
First session on wedding-invite
First build tracked for this project
2.0Mtokens
13prompts
139:51:54time
297lines
Claude Codeclaude-opus-4-8TypeScript$41.18
mostly verifying
Added 5 sample templates with real preview artwork, seeding now works fully offline.
10d ago
Debugged valence-v2
valence-v2
First session on valence-v2
First build tracked for this project
203.7ktokens
6prompts
40:15:51time
15lines
Claude Codeclaude-opus-4-8Markdown$2.42
mostly verifying
12m session running commands and testing. 49K tokens across 4 prompts using claude-opus-4-8.
12d ago
Investigated Gemini API model availability and support
project_litmus
Top 23% session this week
2043 sessions tracked this week across all builders
66.6ktokens
5prompts
9:01:10time
46lines
Claude Codeclaude-fable-5Python$2.48
mostly building
Tested Gemini API models and found version 2.5 is the oldest available.
12d ago
Validated SPARC model outperforms on unseen speaker data
aai
Top 16% session this week
2042 sessions tracked this week across all builders
258.8ktokens
7prompts
5:00:09time
32lines
Claude Codeclaude-fable-5Markdown$7.99
mostly building
Ran experiment on unseen speaker data confirming SPARC outperforms the custom model.