LLM Cost Lens

LLM Cost Lens turns raw provider API logs into attributable spend. It extracts token usage per call and maps it to the task and agent that incurred it, then renders live per-task and per-provider cost across Claude, GPT, and Gemini — so multi-agent spend is measured, not guessed. The extraction-and-normalization logic is encoded in a reusable skill. An internal lab tool that runs inside my own workspace; no public source yet.