发布

  • feat(ai-spend): backend endpoints + diagram/UX refinements

    frostbyte_neo 发布于 2026-06-10 16:26:01 +00:00

    Backend: render spend queries via StringTemplate with log_comment, restore
    UUIDv7 id-range pruning, push leaderboard sort + pagination to ClickHouse (cap
    1000), add output-lane drill-in breakdowns, replace generic filters with a typed
    user_id, model recommendation impact as an enum, and drop unpopulated cost
    fields; expanded endpoint/test coverage.

    Frontend: token-based distribution with a shared TokenCount component, lane
    colors/icons aligned to design, improved single-Claude-Code harness display,
    adaptive cost-breakdown row layout, and home-page bottom padding.

    Co-Authored-By: Claude Opus 4.8 noreply@anthropic.com

    下载附件