expected-findings-template.md

case	<case name, e.g. bq-pipeline | gcp-deploy | sample-plan>
run_at	<UTC ISO timestamp, e.g. 2026-05-01T08:30:00Z>
plan_file	<relative path to plan.md, e.g. examples/sample-plan.md>
models	claude codex gemini cli_version model flags <output of `claude --version`> <as configured in your env> --print cli_version model flags <output of `codex --version`> <as configured> exec --print cli_version model flags <output of `gemini --version`> <as configured> --print
timeout_sec	180
notes	Output is non-deterministic. Findings below are from THIS specific run; your output may differ in wording but should match the pattern.

Expected findings

Replace the metadata header above with your actual run values, then list the consolidated findings below.

Consensus Findings (≥ 2 teams)

#	Finding	Teams	Severity
1	...	Claude, Codex	must-fix

Unique Findings

#	Finding	Team	Severity
2	...	Gemini	should-fix

Apparent Disagreements

• Claude says X; Codex says Y. Both valid; team chooses.

Coverage Gaps

• Dimension N had < 2 concrete scenarios across all teams.

Triple Blind Spot (humans found, all 3 LLMs missed)

• ...