optionality_get_api_usage_stats

io.github.lonniev/optionality-mcp · Verify this server

Aggregated LLM usage and cost for this patron's calls. Returns ``{"models": [{"model", "runs", "total_calls", "total_input_tokens", "total_output_tokens", "total_cost_usd", "priced_runs"}, ...], "totals": {...}}``. One row per distinct model the patron's tool calls have invoked. ``total_cost_usd`` is what the PROVIDER reported for those calls, not a figure reconstructed from a price table — tokens from two models are not comparable money, so a local table goes wrong the moment the route changes model. ``priced_runs`` says how many of ``runs`` carried a reported cost, so a caller can render "$0.42 across 3 of 5 runs" rather than implying the rest were free. ``totals.avg_cost_usd`` is what one call actually costs to serve — the number that says whether a tool's sats price covers its own compute.