@premsai-pendela· Jun 16
Marathon debugging on Naga's Agent
Naga's Agent
MARATHON
31.4x your typical session
2.3M tokens vs typical 71.8K
2.3Mtokens
67prompts
13:59:26time
2,102lines
Codexgpt-5.5Python$52.15
mostly thinking
13h 59m session running commands and testing. 2.2M tokens across 67 prompts using gpt-5.5.
Reconstructed from session history
Tokens / prompt33.7k
Cost / line$0.025
Cache hit97%
Burn rate$3.73/hr
Derived from this session's token and cost data. Not shown on the feed.
Building 32%Verifying 3%Thinking 66%
Comments