Mutation in generator; category H (stop) scoring and judge; category K variants and tool schema variant; evalset plan

- mutation.py: negation mutants, skip WHILE/DO blocks (endless loop blocked RFC ~8 min)
- generator: accept only after mutation check; survivors go back as repair feedback
- runner/judge.py: stop tasks scored 100/30/0; gap by keywords, else judge model
- proxy: tool schema variant generic_v0 (draft); generator make_k_variant
- evalset.py: 88 slots for step F (A-I, H, K)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

harness/agents.py

@@ -43,6 +43,10 @@ class OracleAgent:
     name = "oracle"
 
     def run(self, task, proxy):
+        if task.meta.get("expected_outcome") == "stop":  # the right answer is to stop and name the gap
+            report = "Stopped. Gap: " + task.meta.get("expected_gap", {}).get("description", "")
+            proxy.note("final", report)
+            return report
         for o in task.objects("reference"):
             ident = {"objectType": o["type"], "objectName": o["name"]}
             if o.get("functionGroup"):