Data-driven strategy template: Gather → Classify (rules) → Converge.
Uses a named gatherer to collect data, then applies rule-based classification without LLM synthesis. Suitable for simple threshold/rule actors.
Strategy config shape
%{
"gatherer" => "client_metrics",
"classify_rules" => [
%{"field" => "mrr", "op" => "lt", "value" => 500, "class" => "at_risk", "severity" => "high"},
%{"field" => "last_login_days_ago", "op" => "gt", "value" => 30, "class" => "inactive", "severity" => "medium"}
],
"default_class" => "healthy",
"default_severity" => "low",
"finding_config" => %{
"actor_id_field" => "client_health_actor",
"finding_key_template" => "client:health:${subject_id}",
"subject_kind" => "client",
"subject_id_key" => "client_id"
}
}Rule operators
"lt"— less than"gt"— greater than"eq"— equals"neq"— not equals"in"— value in list"not_in"— value not in list
Rules are evaluated in order. First match wins.