@cbts-narayan-maharjan
#9this weekJoined Jun 15, 2026
73.7MtokensCaptured live
112builds
1108.6hbuild time
35.0klines
7dactive 30d
2wweek streak
Real coding sessions · never self-reported
Highlights
Where the tokens go
53% building · 75.2M classified tokens
Building39.6M53%
Thinking20.0M27%
Delegating8.0M11%
Verifying7.6M10%
15 builds predate classification and aren't included.
2
week streak
12.7M
token record
9d ago
111
prompt record
6mo ago
21h ago
Marathon debugging on chat
chat
DEEP WORK
Biggest session this week
Previous best: 3.4M · +26%
4.2Mtokens
45prompts
1:28:05time
1,998lines
Claude Codeclaude-opus-5-5Python$65.30
mostly building
1h 28m session running commands and testing. 4.2M tokens across 45 prompts using claude-opus-5-5.
21h ago
Debugged Downloads
Downloads
Top 5% build time this week
913 sessions tracked this week across all builders
366.2ktokens
3prompts
7:19:28time
299lines
Claude Codeclaude-opus-5-5Markdown$3.00
mostly building
7h 19m session running commands and testing. 366K tokens across 3 prompts using claude-opus-5-5.
2d ago
One-shot debugging on Downloads
Downloads
ONE-SHOT
Top 3% build time this week
896 sessions tracked this week across all builders
336.9ktokens
1prompts
9:42:44time
118lines
Claude Codeclaude-opus-5-5Python$4.29
mostly building
9h 42m session running commands and testing. 336K tokens across 1 prompts using claude-opus-5-5.
2d ago
Measured extraction accuracy on 1,000 Legal-reviewed contracts
cbtsai-ironclad-ai-tools
LOCKED IN
786.2ktokens
6prompts
21:51:44time
454lines
Claude Codeclaude-opus-5-5HTML$8.25
mostly verifying
Ran the contract extraction on 1,000 Legal-reviewed contracts, split across two frozen test sets.
3d ago
Debugged cbtsai-ironclad-ai-tools
cbtsai-ironclad-ai-tools
DEEP WORK
21.8x your typical session
3.4M tokens vs typical 154.3K
3.4Mtokens
30prompts
27:06:16time
683lines
Claude Codeclaude-opus-5-5Python$57.45
mostly building
27h 6m session running commands and testing. 3.3M tokens across 30 prompts using claude-opus-5-5.
6d ago
Added zoom and draggable panes to the contract review app
cbtsai-ironclad-ai-tools
Top 17% build time this month
5407 sessions tracked this month across all builders
368.3ktokens
5prompts
13:38time
46lines
Claude Codeclaude-opus-5-5Python$4.64
mostly building
Added PDF zoom and draggable split panes to a local contract review tool, then committed the change without pushing.
6d ago
Traced a mislabeled contract back to the workflow entry that caused it
cbtsai-ironclad-ai-tools
Top 19% build time this month
5406 sessions tracked this month across all builders
281.2ktokens
2prompts
8:12time
Claude Codeclaude-opus-5-5$2.27
mostly thinking
Confirmed that a contract's counterparty was entered wrong at launch and flagged its label as known bad.
6d ago
Agent built PDF search and highlight review tool for contract extraction
cbtsai-ironclad-ai-tools
AUTONOMOUS
Biggest cbtsai-ironclad-ai-tools session this week
673.9K tokens
673.9ktokens
0prompts
0:00time
352lines
Claude Codeclaude-opus-5-5Python$10.70
mostly building
Built a local review app that highlights Legal and Claude values on contract PDFs, then measured highlight coverage at 91% across 543 contracts.
7d ago
Agent debugging of Downloads
Downloads
AUTONOMOUS
Top 11% build time this week
467 sessions tracked this week across all builders
178.6ktokens
0prompts
5:30:02time
21lines
Claude Codeclaude-opus-5-5JSON$1.32
mostly building
Autonomous 5h 30m agent session running commands and testing. 178K tokens processed using claude-opus-5-5.
9d ago
Paused idle AWS schedules and held back irreversible deletions
aws-cost-reduction
DEEP WORK
14.2x your typical session
2.0M tokens vs typical 142.3K
3.4Mtokens
40prompts
71:55:14time
Claude Codeclaude-opus-5Markdown$88.80
mostly building
Paused idle dev data schedules with rollbacks saved, and stopped short of deletions until each step was reviewed.
9d ago
One-shot debugging on aws-cost-reduction
duraserv-agent-platform
MARATHON
First $100 session
$149.11 API cost
12.7Mtokens
59prompts
120:50:03time
5,558lines
Claude Codeclaude-opus-5-5Python$149.11
balanced mix
26m session running commands and testing. 416K tokens across 1 prompts using claude-opus-5.
9d ago
One-shot session on scratchpad
cbtsai-analytics-assets
AUTONOMOUS
Biggest session this week
Previous best: 262.4K · +12%
294.6ktokens
0prompts
1:44time
Claude Codeclaude-opus-5-5$2.01
mostly delegating
6s coding session using unknown. 0 tokens across 1 prompts.