Search papers, labs, and topics across Lattice.
This paper introduces a novel knowledge-gated task-construction protocol that distinguishes between task instructions and private conventions necessary for professional agent tasks. By conducting experiments on fifteen calibration tasks, the authors demonstrate that a specific agent configuration achieves a 68.0% pass rate when provided with the artefact containing these conventions, but fails completely without it, highlighting the importance of access to private knowledge. The findings emphasize the need for explicit and testable dependencies on such artefacts to better evaluate agent performance in real-world scenarios.
A single agent configuration shows a stark 68% performance drop when deprived of critical private conventions, underscoring the hidden knowledge gaps in LLM evaluations.
Professional agent tasks often depend on conventions that are absent from public corpora, yet benchmarks rarely control whether an agent has access to those conventions. We introduce a knowledge-gated task-construction protocol that separates a task instruction from a compact artefact containing private conventions, reference tables, and utility operators. Construction-time provenance, byte-identical task instructions across the provided- and withheld-artefact conditions, leak audits, and executable witnesses make dependence on the artefact explicit and testable. Across fifteen calibration tasks, one frontier agent configuration achieves a 68.0% pass rate with the artefact and 0% without it; on one task, a plausible but incorrect artefact also yields 0% across five trials. Deterministic solvers and rule corpora provide exact ground truth for structured tasks, while named criterion-level rubrics support outputs that cannot be checked by a single executable oracle. A configuration-relative calibration screen retains seven tasks satisfying our five-trial empirical knowledge-gating screen. These experiments validate the behavior of the construction protocol; they do not establish that the retained tasks improve post-training. We publicly release part of the task suite and supporting tooling at https://github.com/DatagridsAI/Knowledge-Gated-Task-Construction.
