Overview
Period: 2026-08-04 to 2026-08-05 (last 24h)
Runs analyzed: 12 (12 had token data)
Total tokens: 42,530 across all workflows
Estimated total cost: no estimated_cost/token_usage_summary field was present in the pre-downloaded logs — AIC (Actions-Impact-Cost proxy) totaled 668.18, and total Actions minutes were 43.0
Note: estimated_cost, turns, and token_usage_summary (model/cache breakdown) were not populated in this dataset, so cost, turn-count, and cache-efficiency analysis below are limited to what's available (token_usage, aic, action_minutes).
Workflow Summary
| Workflow |
Runs |
Total Tokens |
Avg Tokens |
Total AIC |
Avg AIC |
Total Action Min |
| Smoke Claude |
12 |
42.5K |
3.5K |
668.18 |
55.68 |
43 |
🔍 Optimization Opportunities
- Smoke Claude — no run exceeds the 100K token/run or $1.00/run thresholds; all 12 runs are well within normal bounds (472–4,450 tokens/run). No immediate action needed on token usage.
- Missing cost/cache instrumentation — none of the 12 runs include
estimated_cost or token_usage_summary (model/cache breakdown). This suggests the api-proxy sidecar's cost/cache tracking may not be fully wired up for this workflow, or the underlying gh aw logs export doesn't populate these fields yet. Recommend verifying api-proxy telemetry export so future reports can include cache read/write and dollar-cost analysis.
- One low-token outlier — run 30964665499 used only 472 tokens vs. ~4,100 typical, and run 30962893347 used 533 tokens. Both completed successfully; likely early-exit/no-op runs rather than a problem, but worth a quick look if this pattern recurs.
Per-Workflow Details
Smoke Claude
- Runs: 12, all
completed/success
- 30978608409 — 4,108 tok
- 30964665499 — 472 tok
- 30962893347 — 533 tok
- 30962536806 — 4,127 tok
- 30960208453 — 4,108 tok
- 30959534937 — 4,450 tok
- 30959146479 — 4,189 tok
- 30956958841 — 4,127 tok
- 30956265669 — 4,117 tok
- 30954778778 — 4,105 tok
- 30928596803 — 4,114 tok
- 30912046220 — 4,080 tok
- Total tokens: 42.5K (avg 3.5K/run)
- Total AIC: 668.18 (avg 55.68/run)
- Total Actions minutes: 43.0 (avg ~3.6 min/run)
- Model breakdown: not available (
token_usage_summary was null for every run)
- Cache analysis: not available in this dataset
- Error rate: 0/12 runs
Workflows Without Token Data
None — all 12 runs reported non-null token_usage.
However, all runs are missing estimated_cost and token_usage_summary, meaning per-run dollar cost and cache read/write breakdown could not be computed. This should be investigated so future reports can surface cache-efficiency and cost-trend insights.
Historical Trend
This is the first Claude token usage report — no prior report issue was found with the claude-token-usage-report label. Future reports will compare against this baseline (42.5K total tokens, 668.18 total AIC, 43 Actions minutes over 12 runs).
Previous Report
None found.
References:
Generated by Daily Claude Token Usage Analyzer · auto · 33.1 AIC · ⊞ 12.9K · ◷
Overview
Period: 2026-08-04 to 2026-08-05 (last 24h)
Runs analyzed: 12 (12 had token data)
Total tokens: 42,530 across all workflows
Estimated total cost: no
estimated_cost/token_usage_summaryfield was present in the pre-downloaded logs — AIC (Actions-Impact-Cost proxy) totaled 668.18, and total Actions minutes were 43.0Note:
estimated_cost,turns, andtoken_usage_summary(model/cache breakdown) were not populated in this dataset, so cost, turn-count, and cache-efficiency analysis below are limited to what's available (token_usage,aic,action_minutes).Workflow Summary
🔍 Optimization Opportunities
estimated_costortoken_usage_summary(model/cache breakdown). This suggests the api-proxy sidecar's cost/cache tracking may not be fully wired up for this workflow, or the underlyinggh aw logsexport doesn't populate these fields yet. Recommend verifying api-proxy telemetry export so future reports can include cache read/write and dollar-cost analysis.Per-Workflow Details
Smoke Claude
completed/successtoken_usage_summarywasnullfor every run)Workflows Without Token Data
None — all 12 runs reported non-null
token_usage.However, all runs are missing
estimated_costandtoken_usage_summary, meaning per-run dollar cost and cache read/write breakdown could not be computed. This should be investigated so future reports can surface cache-efficiency and cost-trend insights.Historical Trend
This is the first Claude token usage report — no prior report issue was found with the
claude-token-usage-reportlabel. Future reports will compare against this baseline (42.5K total tokens, 668.18 total AIC, 43 Actions minutes over 12 runs).Previous Report
None found.
References: