| 1 | #!/usr/bin/env node |
| 2 | // Effort A/B — does CODEGRAPH_OFFLOAD_EFFORT=high improve offload SYNTHESIS FIDELITY vs low? |
| 3 | // Probe-based (no agent): for each repo × effort × rep, run codegraph_explore with the offload |
| 4 | // ON on the canonical question, capture the synthesized answer + AI tokens/cost/latency, then |