Skip to content

armgit / ruflo

Public Imported from GitHub
0

🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated

darwin-core iter 3 — skill-distillation: relax predicate to success && (repeatable || novel) Δ=0.4285

ruv6c1958427a88
diff --git a/scripts/bench-skill-distillation.mjs b/scripts/bench-skill-distillation.mjs
index 5600452c3a..386e635e16 100644
--- a/scripts/bench-skill-distillation.mjs
+++ b/scripts/bench-skill-distillation.mjs
@@ -25,10 +25,13 @@ const TRACES = [
{ id: 't10',success: false, repeatable: true, novel: true },
];
-// Distillation predicate (tick-1 baseline = strict: success && repeatable && novel).
-// Tick-2+ may relax or learn this predicate. The METRIC stays the same.
+// Distillation predicate — tick-3 relaxed: success && (repeatable || novel).
+// Rationale (ADR-155): a successful trace that is either reliably repeatable
+// OR introduces a novel pattern is worth distilling into the skill library.
+// Requiring BOTH (tick-1 baseline) was too strict and dropped reusable wins.
+// The METRIC contract (promoted / successful) is unchanged.
function shouldPromote(trace) {
- return trace.success && trace.repeatable && trace.novel;
+ return trace.success && (trace.repeatable || trace.novel);
}
function run() {