← Files LLM & Agent Builder CopilotARCHIVED FILE
templates/eval-plan.md
497 Bytes · Nov 21, 2026 · 12:41 UTC
# Evaluation plan **Release decision:** **Baseline/candidate (model, prompt, tool schemas):** **Task slices + representative samples:** **Expected outcomes and ground truth:** **Output accuracy / tool selection / tool args / trajectory:** **Side effects + approvals + external state:** **Latency, tokens and cost:** **Grader (deterministic/human/model) and reliability:** **Run reproducibility and sample counts:** **Ship gate and critical fail thresholds:** **Results (not run until measured):**
SHA-256: 0abaded1c8ef66a7005c03fa2184317165b13e00236b459b61e22e461c753574