The limit the operator's attention and judgment place on a human-agent loop — a budget that does not grow with the model, and that grows relatively scarcer as agent capability does. It is AKC's organizing design axis: every phase is shaped to spend the budget only where judgment is irreplaceable. The repository hedges the constraint as a contemporary claim about 2026-era human-agent work, not an eternal law.
Coined by Tatsuya Shimomoto (shimo4228) within the Agent Knowledge Cycle research line (ADR-0010).
The limit the operator's attention and judgment place on a human-agent loop — a budget that does not grow with the model and grows relatively scarcer as agent capability does. AKC's phases are shaped to spend it only where judgment is irreplaceable.
Tatsuya Shimomoto (shimo4228), in the Agent Knowledge Cycle research line (ADR-0010, concept DOI 10.5281/zenodo.19200726).
operator の注意と判断が human-agent ループに課す上限 — モデルとともに成長しない予算であり、エージェント能力が伸びるほど相対的に希少になる。AKC を organize する設計軸であり、すべての phase は、判断が代替不能な場所にだけこの予算を使うよう形づくられている。repo はこの制約を「2026 年時点の human-agent work についての contemporary な主張であって、永遠の法則ではない」と hedge している。