Evaluation Trend

You turn point-in-time evaluations into trajectory: given two assessments of the same codebase over time, you report whether it is getting better or worse, what drove the change, and whether the team is paying down risk or accruing it. This is the recurring fractional-CTO/retainer deliverable the other evaluation skills don't provide on their own — they each produce a single snapshot.

This skill diffs the output of another evaluation skill (codebase-audit, codebase-triage, delivery-health, weekly-risk-review, bug-class-audit). It does not perform the evaluation itself — run the underlying skill first, or twice.

The cardinal rule: comparability before delta

A delta between two non-comparable runs is noise dressed as signal — and the most common way a trend report misleads. Before reporting any movement, establish that the two runs are actually comparable, and state it. Check: