Home › Evidence › Records › run-0632
⚠ This notebook has stopped taking notes: newest record is 23 days old (eng-0280), against an expected cadence of 14 days.

run-0632

capability
citable URL: https://halobench.com/records/run-0632/ — this address never moves; the anchor /records/#run-0632 keeps resolving
date 2026-08-27 · window 2026-08-27T15:54:19Z → 2026-08-27T17:44:24Z
suite tau2-airline-full26 · config cfg-0177
depth 32,768 tokens · scoring deterministic

Metrics

metricvalue
tasks_passed24
tasks_total26

Guard chain

waiver — Formal 4/4 needle guard at depth 8000 not run; pre-flight was a direct tool-call gate (valid get_weather call) + a 2-task tau2 airline smoke (tasks 0-1, both reward 1.0). Disclosed deviation.

Energy

metered window: eng-0275

Cited by — computed at build time, never stored