Expand description
2026-09-26: SSM state poisoning gate: driver runs it, probe holds the
script, compare judges each replay round, score decides, toolcall
checks path-independence of tool calls, and report renders.
Owner: bench, SSM poisoning gate. Invariants: none beyond the types.
Re-exports§
pub use driver::DESCRIPTOR;pub use driver::METADATA;
Modules§
- compare
- 2026-09-26: Round comparison: every replay round against the reference round.
- driver
- 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.
- probe
- 2026-09-26: The probe: a fixed conversation script that every round replays from scratch against the same server, and the anchors the reference round must meet.
- report
- 2026-09-26: How a poisoning run is presented: pure functions of the
super::score::Score. - score
- 2026-09-26: The decision logic: collected round records → a
Score→ the replay verdict. - toolcall
- 2026-09-26: Path-independence of tool calls: one target turn, whose correct answer is to call nothing, reached by three predecessor paths that share the same long prefix. The replay script offers no tools, and its comparison passes reworded prose, so it cannot see a tool call that moves with history.