Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use CodexBar CLI local cost usage to summarize per-model usage for Codex or Claude, including the current (most recent) model or a full model breakdown. Trigger when asked for model-level usage/cost data from codexbar, or when you need a scriptable per-model summary from codexbar cost JSON.
.claude/skills/model-usage/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -73% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -45% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -62% | 0% |
Get per-model usage cost from CodexBar's local cost logs. Supports "current model" (most recent daily entry) or "all models" summaries for Codex or Claude.
TODO: add Linux CLI support guidance once CodexBar CLI install path is documented for Linux.
1) Fetch cost JSON via CodexBar CLI or pass a JSON file. 2) Use the bundled script to summarize by model.
bashpython {baseDir}/scripts/model_usage.py --provider codex --mode current python {baseDir}/scripts/model_usage.py --provider codex --mode all python {baseDir}/scripts/model_usage.py --provider claude --mode all --format json --pretty
modelBreakdowns.modelsUsed when breakdowns are missing.--model <name> when you need a specific model.codexbar cost --format json --provider <codex|claude>.bashcodexbar cost --provider codex --format json > /tmp/cost.json python {baseDir}/scripts/model_usage.py --input /tmp/cost.json --mode all cat /tmp/cost.json | python {baseDir}/scripts/model_usage.py --input - --mode current
--format json --pretty).references/codexbar-cli.md for CLI flags and cost JSON fields.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 2,703 | 5,631 | +108% | 1 | 1 | 0% | 412 | 682 | +66% | 0 | 0 | — |
case-02 | fail→pass | 13,500 | 2,770 | -79% | 1 | 1 | 0% | 3,128 | 844 | -73% | 0 | 0 | — |
case-03 | fail→pass | 6,820 | 1,581 | -77% | 1 | 1 | 0% | 1,316 | 658 | -50% | 0 | 0 | — |
case-04 | pass→pass | 8,628 | 2,686 | -69% | 1 | 1 | 0% | 1,665 | 823 | -51% | 0 | 0 | — |
case-05 | fail→pass | 5,710 | 4,115 | -28% | 1 | 1 | 0% | 1,033 | 726 | -30% | 0 | 0 | — |
case-06 | fail→pass | 7,223 | 2,109 | -71% | 1 | 1 | 0% | 1,238 | 683 | -45% | 0 | 0 | — |
case-07 | pass→pass | 11,005 | 1,974 | -82% | 1 | 1 | 0% | 1,962 | 698 | -64% | 0 | 0 | — |
case-08 | fail→pass | 10,182 | 1,862 | -82% | 1 | 1 | 0% | 1,828 | 698 | -62% | 0 | 0 | — |
case-09 | fail→pass | 12,231 | 3,100 | -75% | 1 | 1 | 0% | 2,149 | 965 | -55% | 0 | 0 | — |
case-10 | pass→pass | 10,382 | 1,892 | -82% | 1 | 1 | 0% | 1,749 | 646 | -63% | 0 | 0 | — |
case-11 | fail→pass | 7,997 | 1,540 | -81% | 1 | 1 | 0% | 1,572 | 633 | -60% | 0 | 0 | — |
case-12 | fail→fail | 5,421 | 2,308 | -57% | 1 | 1 | 0% | 879 | 779 | -11% | 0 | 0 | — |
case-13 | fail→pass | 7,221 | 2,166 | -70% | 1 | 1 | 0% | 1,192 | 730 | -39% | 0 | 0 | — |
case-14 | fail→pass | 7,043 | 1,612 | -77% | 1 | 1 | 0% | 1,253 | 651 | -48% | 0 | 0 | — |
case-15 | fail→pass | 12,277 | 2,653 | -78% | 1 | 1 | 0% | 1,914 | 814 | -57% | 0 | 0 | — |
case-16 | pass→pass | 4,627 | 1,834 | -60% | 1 | 1 | 0% | 727 | 671 | -8% | 0 | 0 | — |
case-17 | fail→fail | 3,661 | 1,947 | -47% | 1 | 1 | 0% | 613 | 746 | +22% | 0 | 0 | — |
case-18 | fail→fail | 10,416 | 1,959 | -81% | 1 | 1 | 0% | 1,844 | 729 | -60% | 0 | 0 | — |
case-19 | fail→pass | 7,272 | 3,609 | -50% | 1 | 1 | 0% | 1,141 | 906 | -21% | 0 | 0 | — |
case-20 | fail→pass | 13,736 | 4,080 | -70% | 1 | 1 | 0% | 2,876 | 1,155 | -60% | 0 | 0 | — |
case-21 | fail→fail | 12,370 | 8,230 | -33% | 1 | 1 | 0% | 2,143 | 1,839 | -14% | 0 | 0 | — |
case-22 | pass→pass | 8,704 | 2,798 | -68% | 1 | 1 | 0% | 1,396 | 928 | -34% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +55 percentage points is the difference between those two pass rates over the 21 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.