https://southbase.dev Writing on the hard problems in agent observability. en https://southbase.dev/blog/four-token-buckets/ https://southbase.dev/blog/four-token-buckets/ Collapsing Claude's four billing rates into "input" and "output" does not just lose detail — it reports the wrong cost and hides the one efficiency lever you have. Sat, 08 Aug 2026 00:00:00 GMT https://southbase.dev/blog/why-llm-judges-are-distrusted/ https://southbase.dev/blog/why-llm-judges-are-distrusted/ The failure modes of LLM-as-judge are specific and documented. More metrics do not fix them — calibration does, and almost no tool ships it. Thu, 06 Aug 2026 00:00:00 GMT https://southbase.dev/blog/span-trees-break-for-agents/ https://southbase.dev/blog/span-trees-break-for-agents/ The span tree is the right shape for a request and the wrong shape for a ten-minute agent run. Here is what actually goes wrong, and what replaces it. Tue, 04 Aug 2026 00:00:00 GMT