Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →lingtai-ai/token-usage
Nested swiss-knife reference for token usage, cost, cache, and tool-call/API-call reports.
Passed 16 of 19 test cases with the skill, and 5 of 19 without it.
and 15 more in the full report
13 fixed5 passed either way3 still failing1 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 13 test cases the skill fixed. Same model, same prompt, with and without the skill.
What one-line shell python command can be run to verify whether the pricing engine package is installed in the current environment?
| Checked | Without skill | With skill |
|---|---|---|
| Uses python3 -c "import litellm" | Missed | Passed |
The instructions your model reads when the skill loads.
Network-wide token cost analysis powered by litellm's model pricing database (2700+ models), plus runtime-event trend analysis for tool calls per API call.
Run the bundled cost script:
~/.lingtai-tui/runtime/venv/bin/python3 ~/.lingtai-tui/utilities/swiss-knife/reference/token-usage/scripts/cost_report.py /path/to/.lingtai
Cost-report optional flags:
--json — output as JSON instead of table--by-model — group by model instead of by agent--since YYYY-MM-DD — only count entries after this date--top N — show only top N agents (default: all)--custom-pricing FILE — load custom pricing overrides from JSONUse this when the human asks for “tool calls per API call”, “tools per API call”, “agentic intensity”, or how tool-heavy the local LingTai network has been over time. The script reads LingTai logs/events.jsonl files, not token_ledger.jsonl, because it needs runtime event IDs.
~/.lingtai-tui/runtime/venv/bin/python3 ~/.lingtai-tui/utilities/swiss-knife/reference/token-usage/scripts/tool_calls_per_api_call_trend.py \
/path/to/project/.lingtai \
--days 5 \
--timezone America/Los_Angeles \
--model gpt-5.5
Optional flags:
--out-prefix PREFIX — write PREFIX.md, PREFIX.json, and PREFIX.csv--model MODEL — also emit a model-specific table for an exact model name; repeatable--include-daemons — include daemon event logs instead of excluding themMetric definitions:
type == "llm_response"), deduplicated by (events log path, api_call_id).test cases passed, vs 5 / 19 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.