Evaluation receipt / invocation-doctor

Invocation Doctor

Two valid pairs tied, and a third pair was invalidated when both arms missed the objective gate.

Inconclusive
2Valid pairs
2Blind ties
2/3Treatment verifier passes
1Invalid pair

What this proves

Completed the trigger-repair task without showing a measurable advantage.

Status

inconclusive. No lift was demonstrated over the empty baseline.

Method

Three attempted paired Harbor runs, with invalid pairs excluded before winner aggregation.

Model

openai/gpt-5.6-sol

Limitations
  • One representative trigger-repair task
  • One model
  • Two valid paired samples
Exact packagesha256:ecae0aed50f012b3534e811c901e6f86915444b768b948592f84f99140324f1e