https://session-viz.com/blog/ Field notes on measuring coding agents. en-GB https://session-viz.com/blog/an-agent-that-edited-225-files-and-reported-nothing/ https://session-viz.com/blog/an-agent-that-edited-225-files-and-reported-nothing/ Sat, 15 Aug 2026 00:00:00 GMT Cursor renamed edit_file to edit_file_v2, the adapter kept exact-match lookup, and 274 sessions were filed as never having tried to write anything. https://session-viz.com/blog/model-comparisons-are-confounded/ https://session-viz.com/blog/model-comparisons-are-confounded/ Sat, 15 Aug 2026 00:00:00 GMT Eight models cleared the volume floor, giving 28 pairs. Twenty-five never overlapped, two were underpowered, and the one testable pair came out null. https://session-viz.com/blog/reading-cursors-chat-history/ https://session-viz.com/blog/reading-cursors-chat-history/ Sat, 15 Aug 2026 00:00:00 GMT Cursor keeps every conversation in one 4.4 GB SQLite file and records no working directory, so the repository has to be reconstructed from file URIs. https://session-viz.com/blog/where-the-tokens-go/ https://session-viz.com/blog/where-the-tokens-go/ Sat, 15 Aug 2026 00:00:00 GMT Across 2,364 agent runs, cache-read was 96% of all tokens and output was 0.39%. Eight runs account for half the cache-read; one accounts for 22%.