Understanding Check Grades

Overview

axm init_check scores your project against checks derived from the project template and CI configurations. Papers are validated by axm-lab's paper_check; Init returns an explicit routing error for paper contexts.

Grade Scale

Grade	Score Range	Meaning
A 🏆	90–100	Meets at least 90% of selected weighted checks
B ✅	75–89	Good — minor improvements needed
C ⚠️	60–74	Acceptable — several gaps
D 🔧	40–59	Below standard — significant work needed
F ❌	0–39	Failing — major structural issues

Scoring System

Each check has a weight (1–5 points). The score is computed over the checks that actually run for the project's context, not against a fixed point total:

Text Only

Score = round(earned points / weight of executed checks × 100)

The denominator is dynamic. The check engine selects which checks run from the project context (standalone, workspace, member, paper, experiment). A workspace selects workspace checks and skips package-only checks. For members, some checks are skipped while CI and shared tooling checks are redirected to the workspace root. Paper validation is owned by axm-lab through paper_check. Experiment checks are owned by axm-lab; Forge returns migration guidance instead of a score. See project contexts.

If no weighted check applies, the result is N/A: structured score and grade are null and the CLI exits successfully. Otherwise the score is normalized to 0–100 and mapped to the grade boundaries above.

Check catalogue

The Python catalogue lists each check and weight, plus the framework-specific category sets. Scores are based on the selected checks, not on the size of that catalogue.

Improving Your Score

Every failed check includes a Fix instruction telling you exactly what to do. Run axm init_check iteratively until you reach Grade A.

Quick win

Scaffolds aim at the configured standard; measure the actual result after installing dependencies and hooks. Grade A is not the CLI success threshold: an applicable run exits successfully only at 100/100.