Scientific agents can retrieve relevant passages but often lose functional order, mix evidence across sources, or state conclusions that go beyond the retrieved record. To address these issues we introduce PathAnchor, a bounded scientific reasoning system built on path‑structured evidence workspaces. Instead of treating passages or extracted concepts as independent units, the system retrieves source‑linked Material‑Sensor‑Signal‑System trajectories that preserve role, direction, and the evidence supporting each transition. A controller employs three read‑only tools: searching paper‑specific trajectories, tracing paths across candidate sources, and opening exact evidence, then produces a claim‑cited answer with an explicit evidence boundary.
On 120 single‑ and cross‑paper flexible‑sensor questions, PathAnchor achieves an accuracy of 82.6%, leading six evaluated systems. Under a matched controller, corpus, and six‑call budget, replacing unordered concept graphs with path‑structured records raises source recall from 61.3% to 82.9%, increases the proportion of answers whose claims all cite opened evidence from 69.2% to 90.0%, and reduces tool calls. These results demonstrate that evidence organization significantly impacts retrieval and citation completeness under fixed agent resources.
Review