A 7B repo scout matches the best frontier fixer on SWE-bench Pro at one-fifth the cost, and the router turns out not to matter.

arXiv 2608.04804 sends a 7B searcher into the repo first, sandbox-verifies its reproduction claims and strips false ones, then routes to one of four frontier fixers. On the full 266-task Python slice under the official capped budget it solves 159 vs 158 for the best single model. The honest ablation is the finding: always using the cheapest fixer with the handoff ties the routed system. The verified context handoff carries the result, not the routing. Calibration suggests it redistributes solving ability upward for cheap models while slightly hurting the strongest.