from datus-agent1,161
Interactively define MetricFlow metrics from natural-language business descriptions; proposes, validates, and dry-runs metric YAML for semantic modeling.
This skill guides an agent through defining MetricFlow metrics from plain-English business requests. It analyzes existing metrics, inspects table structure or reference SQL, proposes core metric candidates, and produces validated metric YAML plus dry-run SQL. The workflow enforces user confirmations before generating models, ensures semantic models exist, and validates generated metrics so they can be synced to the knowledge base.
Use this when you need to convert product/business questions ("total revenue", "conversion rate", "orders per day") into production-ready MetricFlow metrics, or when you must analyze multiple SQL examples and deduplicate aggregation patterns. Ideal for teams onboarding metrics or migrating ad-hoc SQL into an explicit semantic layer.
Best used by agents that can run SQL inspection and file operations (CLI-capable LLM agents, MetricFlow-adapter orchestrators).
MetricFlow metric definition skill for the Datus agent ecosystem. Provides a thorough 5-phase interactive workflow for defining MetricFlow metrics from natural language, with detailed rules for metric type detection, semantic model creation, and validation. No scripts included — purely instructional SKILL.md. No security concerns: no shell commands, no network calls, no credential handling. Quality is good but the skill is extremely long and monolithic, mixing detailed YAML format rules with high-level workflow. Architecture is decent with clear phase separation but all content lives in one massive SKILL.md with no references/ directory or scripts/. Usefulness is niche — requires the Datus agent ecosystem and MetricFlow tooling, limiting audience to MetricFlow/dbt users already on that platform.
Pure instructional skill, no executable code. Well-structured phases but overly verbose. Could improve architecture by extracting YAML format rules and common pitfalls into references/. Niche audience due to Datus/MetricFlow dependency.
