run-scenariolisted

# Run a Scenario One protocol on both ends of the ODD loop: the same commands that produced the numbers in an observation report produce the numbers that verify the fix. A scenario that cannot be replayed verbatim makes before/after comparison an impression, not a measurement. **A step this package ships a script for is run, never rewritten.** The ad-hoc drive is this skill's `scripts/drive_scenario.py` (step 2); a stored benchmark's replay is `k6-guides`' `scripts/replay_benchmark.py` (`references/benchmark-replay.md`). A curl loop, a helper with a `sleep` in it, a poller written for either yields a different command each time — two runs then measure two things. ## Read by situation Steps 1 to 5 and the rules are the method; a reference is read by the block that applies, never whole: | Situation | Reference | | --- | --- | | Every drive: the clean-base order and the reset decisions, the identity the queries are qualified by and the instance read from the rows, the run's t0 after the warmup — a port already served, a remote target the run cannot launch, a restart that is not possible, a reset that is forbidden | [references/run-identity.md](references/run-identity.md), the block that applies | | An iteration that is expensive or non-deterministic; the watch of a run someone else drives, on a backend that ships no watch script | [references/long-scenarios.md](references/long-scenarios.md) | | A stored k6 benchmark under `.odd/benchmarks/<name>/` — driven here, or driven e