@spec agentic_classification(float()) :: {:ok, String.t()}

agentic_classification/1: returns classification label for agentic contribution ratio. Ratio represents the fraction of commits from bots or AI-assisted contributors. Returns "human", "mixed", or "agent".

@spec commit_change_size_risk(non_neg_integer()) :: {:ok, String.t()}

last_commit_size_risk/1: returns a text enumeration for the risk based on the size of the last commit

@spec commit_currency_risk(non_neg_integer()) :: {:ok, String.t()}

commit_currency_risk/1: returns text enumeration for commit currency risk

@spec contributor_risk(non_neg_integer()) :: {:ok, String.t()}

contributor_risk/1: returns text enumeration for count

@spec functional_commit_currency_risk(non_neg_integer()) :: {:ok, String.t()}

functional_commit_currency_risk/1: the risk from time since the last commit that carried information (#244).

Configured separately from commit_currency_risk/1 on purpose. The last substantive commit is always at or before the last commit of any kind, so under shared thresholds this metric could only ever be the worse of the two -- and since the report's verdict is a maximum, the plain metric could never be the one that decides. Tuning one would silently move the other. These are the levels that actually govern the verdict; commit_currency_* is kept as it was so the two remain comparable in a report.

The fallbacks below are deliberately distinct from each other. Where commit_currency_risk/1 falls back to 52 for both its high and critical levels -- making "high" unreachable for a library consumer who configures nothing -- every level here is reachable without configuration.

@spec functional_contributors_risk(non_neg_integer()) :: {:ok, String.t()}

functional_contributors_risk/1: returns the enumerated risk based on input contributors list

@spec sbom_risk() :: {:ok, String.t()}

sbom_risk/1: returns text risk level for identified sbom_risk