/evaluate

Creates a vendor evaluation framework and scores proposals against project requirements, architecture principles, and customizable templates.

Evaluates a developer tool against structured criteria, scoring functionality, performance, and other dimensions. Produces a weighted score, strengths/weaknesses summary, and a recommendation with confidence level.

Runs a heuristic evaluation of a UI design, identifying usability issues against Nielsen's heuristics, analyzing user flows, and checking accessibility, then outputs a prioritized report with improvement recommendations.

Evaluates a factor library's out-of-sample IC, ICIR, and decay by recomputing metrics on the test split. Supports --period both to surface train→test decay.

Evaluates front-end UI code for usability issues using 15 heuristics, rates severity 0-4, and produces a structured report with summary, findings, strengths, and recommendations. Accepts file/dir paths.

Evaluates web artifacts (URL, HTML, screenshot, copy) with Perception-First Design corpus, producing layered scores, citations, and rebuilt HTML fixes for top issues.