# Post-trial payload exposure fix

The completed six-call trial is frozen. This offline change adds
`local_packet_validation` only to a successful `promo_incrementality` result.
It exposes checks that the tool had already performed on the supplied packet,
while keeping supplied experiment declarations separate from facts that require
external authentication or real protocol verification.

The C09 frozen input and reference card were recomputed after the change.
`DECISION` (`ACT`), `ACTION` (`bounded_pilot`), and every `P&L IMPACT` value
match the frozen reference; contribution delta remains CNY 5767.4.

Offline validation passed: 26 tests in `tests.test_sdk_agent` and
`tests.test_tools`. The SDK evidence view now exposes the successful packet
checks, and tests confirm that baseline contamination and cross-channel
assignment crossover still withhold the estimate.

No provider request or real-model call was made for this fix. It does not show
that either role wording, model quality, business benefit, ROI, or real-world
promotion-protocol performance improved.