BENCHMARK APPLICABILITY WORKSHEET
Blank planning record — not a benchmark result, recommendation, approval, or completed test

DECISION
Decision to support:
Decision owner:
Consequence of a poor choice:
Decision deadline:

WORKLOAD
Task description:
Input modalities:
Output form:
Output length or context limits:
Languages:
Required tools:
Prohibited tools or data flows:
Quality attributes that matter:
Failure costs:

PUBLISHED BENCHMARK
Publisher:
Benchmark or index name:
Exact version/date:
Methodology URL:
Task subset relevant to this decision:
Harness and prompt setup:
Model/provider configuration:
Reasoning mode and temperature/settings:
Repeat count:
Grader or verifier:
Scoring and aggregation rule:
Known limitations stated by publisher:
Reported uncertainty or confidence treatment, if any:
Cost measurement boundary, if reported:
Latency measurement boundary, if reported:

APPLICABILITY
Matching workload properties:
Material mismatches:
What the published result can support:
What it cannot support:
Trigger for rechecking this record (method/version/workload change):

LOCAL CONFIRMATION — PROPOSED, NOT EXECUTED
Smallest remaining question:
Authorized harmless inputs:
Frozen answer key or review rubric:
Exact runtime/configuration to record:
Repetitions needed and why:
Human review required:
Stop conditions:

EVIDENCE STATUS
Published methodology reviewed: unreviewed
Benchmark version recorded: unreviewed
Applicability reviewed: unreviewed
Local confirmation approved: unreviewed
Local confirmation executed: no
Decision made: no

Notes: