https://usekibble.com/blog/ Notes on measuring what AI coding agents cost a software team: usage logs, spend per engineer, estimated against billed. en https://usekibble.com/blog/claude-code-cost-per-engineer https://usekibble.com/blog/claude-code-cost-per-engineer Sat, 29 Aug 2026 00:00:00 GMT Three ways to break one Anthropic invoice into spend per engineer, team, model and repo: a local usage tool, a proxy, or a collector that reads the logs the agent already writes and sends counts only. What each one sees, and what it costs you in privacy. https://usekibble.com/blog/where-coding-agents-keep-their-usage-logs https://usekibble.com/blog/where-coding-agents-keep-their-usage-logs Sat, 29 Aug 2026 00:00:00 GMT The JSONL transcripts under ~/.claude/projects and ~/.codex/sessions, the four token fields in them, why cache reads dominate the tokens but not the cost, and which fields a collector should never send. https://usekibble.com/blog/estimated-vs-billed https://usekibble.com/blog/estimated-vs-billed Sat, 29 Aug 2026 00:00:00 GMT Local logs give you tokens at list price; the vendor's Admin API gives you what was charged. They disagree for six good reasons. The rule that keeps a dashboard honest: label both, let billed win the day, never sum them.