cost-report

To generate an inference cost report, the skill reads memory/token-usage.csv, computes costs using per-field pricing, and aggregates the data by skill and time windows to output memory/inference-cost.md.

Calculating per-skill token costs from a CSV involves applying per-field pricing to the usage data, aggregating the results by skill, and generating a breakdown of expenses for 7-day and 30-day windows.

Daily burn estimation works by aggregating memory/token-usage.csv data across runs, computing costs with per-field pricing, and logging the daily token consumption rate into memory/inference-cost.md.

Yes, the skill processes usage logs to produce optimization signals that highlight heavy-usage skills and suggest cadence adjustments, helping you trim unnecessary token usage and manage inference costs.

You need a memory/token-usage.csv file containing token usage data. The skill uses this CSV to compute per-field pricing, aggregate costs by skill, and write the final breakdown to memory/inference-cost.md.

You should use an automated cost report when you need to track daily burn estimates, identify heavy-usage skills, and receive optimization signals to guide cadence decisions for reducing inference costs.