A builder tracks coding-agent cost per feature by token usage and git commits
andfanilo · x · 2026-07-27
A builder shares a workflow for estimating feature implementation cost by token consumption instead of man-day planning.
Using ccusage, they linked API usage to git commits to break down per-feature costs and say they were “vibecoding” with real API spend visibility.
Related event: Estimating Feature Costs via Token Usage(2 posts)→
More from coding & agent
- Start simple, then add memory and loops as your agent problem becomes clear — bendee983 · 2026-07-27
- TB2-Fn shows agent benchmark scores can swing 40% from verifier errors — abeirami · 2026-07-27
- NVIDIA’s NOOA turns AI agents into plain Python classes — s_scardapane · 2026-07-27
- Linus brings OpenAI- and Anthropic-compatible AI orchestration to Apple Silicon — dee_hw · 2026-07-27
- Two open-source tools turn terminals and home AI fleets into managed desktops — dee_hw · 2026-07-27
- Honeycomb-lab manages a home AI fleet with live maps and auto failover — dee_hw · 2026-07-27