CLI Reference
featuresmith score
Understanding ML Readiness Score CLI integration
Overview
There is no standalone featuresmith score CLI subcommand. Instead, the versioned, explainable ML Readiness Score is computed automatically and displayed inline as a section at the end of the standard featuresmith review report.
How it is Used
By default, running a review includes the scorecard output. To prevent calculating or rendering the score scorecard (useful for clean log outputs or minimal JSON size), call the review command with the --no-score flag.
featuresmith review train.csv --no-scoreConsole Scorecard Example
When computed, the CLI console renderer formats the score and displays contributing issues dynamically:
ML Readiness Score (scoring v0.3.0)Overall: 98.1/100 Schema Health: 100/100 Missing Values: 85/100 (1 finding(s)) Feature Quality: 100/100 Distribution Health: 100/100 Leakage Risk: 100/100 Data Quality: 100/100 Consistency: 100/100Summary: Overall ML Readiness is 98.1/100 across 7 dimension(s); 6 fully healthy, 1 with findings lowering the score.What would improve this score: - Address the flagged issue: High missing values in column 'age' (in column 'age').JSON Integration
If the review is output with --format json, the score object is nested as the score field in the output payload. If --no-score is requested, the field yields null rather than 0, keeping "not scored" distinct from "scored poorly".
Related Documentation
See the SDK equivalent fs.score() and the review CLI reference featuresmith review.