Usage
{% if message %}{{ message }}
{% endif %} {% if error %}{{ error }}
{% endif %}Tokens read from the Claude Code transcripts the hooks report. relaybay knows neither prices nor quotas. Subagents count to their session. "unknown" means the data does not say.
{% if not usage_enabled %}Reading is off (usage_enabled = false): this is what was read before.
{{ "{:,}".format(report.unknown_lines) }} transcript line(s) were not understood and are not counted.
{% endif %} {% if report.capped_files %}{{ report.capped_files }} subagent transcript(s) over the limit per session are not counted.
{% endif %} {% if report.unread_files %}{{ report.unread_files }} transcript file(s) could not be read at the last pass.
{% endif %}Budgets
Tokens as in the total above, cache reads included. The day is the UTC day. A budget warns at 80% and at 100%; only one set to Stop agents stops them, at their next tool call, and only you release that. Budgets lag: they are checked after each read of the transcripts, at least once a minute. A session on several tasks counts in full on each; a task counts from the moment a session was bound to it, from version 0.8 on.
{% if not usage_enabled %}Budgets are blind: reading is off (usage_enabled = false), so no new tokens count against them.
No usage for this filter.
Usage appears after the Claude Code hooks of a session report its transcript ({{ report.files }} file(s) tracked so far); the reader looks at them in the background.
| {{ "Day" if report.group == "day" else "Week of" }} | Session | Project | Model | {{ headers() }}
|---|---|---|---|
| {{ row.period }} | {% if row.session_id is not none %}{{ row.session }}{% else %}{{ row.session }}{% endif %} | {{ row.project }} | {{ row.model }} |
{{ numbers(row.total) }}
| Total | {{ numbers(report.total) }}|||
Showing the newest {{ shown_rows|length }} of {{ report.rows|length }} rows; the total counts them all. Narrow the filter to see the rest.
{% endif %} {% endif %}