speckit-token-analyzer-compare

To compare token usage, evaluate a baseline run against a variant run to compute metric deltas for input tokens, output tokens, total tokens, cache reads, and quality scores. This comparison reveals whether the variant improved or worsened efficiency.

Baseline-versus-variant evaluation is the process of measuring token consumption differences between two spec-kit analysis runs. It calculates efficiency ratios and percentage deltas to support accept, reject, or review decisions for workflow changes.

Yes, you need a spec-kit project structure with existing token analysis summaries. The comparison requires saved baseline and variant runs to calculate per-command breakdowns and generate a markdown comparison report.

Calculate efficiency deltas by comparing the token consumption metrics of a baseline run against a variant. Compute the percentage change across input, output, and total tokens to determine if the new approach reduced usage.

The best way to decide is to generate a markdown comparison report evaluating efficiency ratios and percentage deltas between the baseline and variant runs. Review the quality scores and token metrics to determine adoption.

Token analysis comparison reports are saved as markdown files in the token analysis comparisons directory within the spec-kit project structure, documenting the metric deltas and accept or reject recommendations.