Expand description
2026-09-26: The SSM state poisoning gate’s driver: replay the probe script against one server, compare every replay with round 0, then run the tool-call path probe.
Owner: bench, SSM poisoning gate. Invariants:
- A replay round that errors is scored
super::compare::RoundVerdict::Unmeasured; the run goes on. - The run fails on a reference round that misses the script anchors, on a short run, on any collapsed, unmeasured or zero-cache replay, and on a tool call that differs between predecessor paths.
Structs§
Constants§
- DEFAULT_
ROUNDS - 2026-09-26: The default replay count. The gate’s BENCH.toml entry bounds the
recorded
roundsmetric at exactly 12. - DESCRIPTOR
- METADATA