teamchong a19ce30a48 feat(core): quantize history collapse boundary onto a fixed grid (#28) ...

Snap the collapse cutoff to a `collapseChunk`-sized grid (default 50) so
the rendered history image stays byte-identical for a full window of
turns. That lets Anthropic's prompt cache `cache_read` the history
prefix (0.1x) instead of re-billing `cache_create` (1.25x) every turn.
The cutoff floors at `minCollapsePrefix` (clamped to the raw cutoff) so
short conversations still collapse without the boundary drifting.

Also log `history_image_sha8` per request -- a sha8 of the concatenated
history-image base64. An unchanged hash across consecutive collapsed
events is ground-truth proof the prompt cache is being hit; a hash that
moves every turn signals the cache-key drift regression is back.

2026-05-21 21:43:22 -04:00

..

fixtures

docs(render): clarify 5x8 atlas sizing

2026-05-21 17:45:35 -04:00

dashboard-api.test.ts

dashboard: sessions UI + stats panel + smooth updates + cleanup

2026-05-18 22:14:00 -04:00

history.test.ts

feat(core): quantize history collapse boundary onto a fixed grid (#28)

2026-05-21 21:43:22 -04:00

paging.test.ts

docs(render): clarify 5x8 atlas sizing

2026-05-21 17:45:35 -04:00

proxy-usage.test.ts

feat(proxy): SSE+JSON output measurement scanner

2026-05-20 15:53:58 -04:00

public-api.test.ts

fix(measurement): pair orphan tool_use blocks in cacheable-prefix probe

2026-05-21 15:13:23 -04:00

render.test.ts

docs(render): clarify 5x8 atlas sizing

2026-05-21 17:45:35 -04:00

restart.test.sh

refactor: remove all CLI flags and behavior env vars — single codepath

2026-05-19 17:01:05 -04:00

sessions.test.ts

feat(dashboard): real saved% from count_tokens, rip α-regression dead code

2026-05-19 22:00:14 -04:00

stats.test.ts

cli: add 'pixelpipe stats' subcommand for offline log analysis

2026-05-18 16:13:04 -04:00

tracker.test.ts

feat(transform): per-bucket char attribution telemetry (Task #18 Phase 1)

2026-05-20 21:21:34 -04:00