```json { "table_name": "nia_message_llm_calls", "table_type_tags": [ {"tag": "event", "confidence": 0.90}, {"tag": "path", "confidence": 0.25} ], "primary_type": "event", "insufficient_evidence": false, "entity": { "columns": ["message_id"], "unique_per_row": false }, "time_axis": { "columns": ["id"], "kind": "sequence" }, "reasoning": "The entity message_id repeats (RPUV 7.34, Uniqueness 0.136) along a sequence axis — id is a monotonically increasing call counter (Monotonicity 1.000, Uniqueness 1.000). Each row records an LLM API call occurrence: a message triggers multiple calls with varying tokens_in/tokens_out/cost_usd/latency_ms. The rows are bare occurrences to count and analyze; replaying them does not reconstruct a meaningful message state (no balance or holding being mutated) → event. The weak path signal comes from tokens accumulating across calls, but the values are per-call absolutes, not deltas." } ```