{"paper":{"title":"PrecisionCUA: Iterative Visual Refinement for Pixel-Precise Cursor Grounding in Code Editors","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"Multi-turn refinement using visual feedback from prior attempts achieves higher click precision and task success in GUI grounding for dense coding interfaces than single-shot prediction.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Gaurav Mittal, Himangi Mittal, Nelson Daniel Troncoso, Yu Hu","submitted_at":"2026-04-14T17:55:46Z","abstract_excerpt":"Computer Use Agents (CUAs) fundamentally rely on graphical user interface (GUI) grounding to translate language instructions into executable screen actions, but editing-level grounding in dense coding interfaces (such as VS Code and Cursor), where sub-pixel accuracy is required to interact with dense IDE elements, remains underexplored. Existing approaches typically rely on single-shot coordinate prediction, which lacks a mechanism for error correction and often fails in high-density interfaces. In this technical report, we conduct an empirical study of pixel-precise cursor localization in cod"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"multi-turn refinement significantly outperforms state-of-the-art single-shot models in both click precision and overall task success rate.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That models can reliably interpret and act on visual feedback from prior attempts to self-correct without introducing additional errors or failing to adapt to dynamic UI changes.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Multi-turn visual feedback refinement outperforms single-shot coordinate prediction for pixel-precise GUI grounding in complex coding environments.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Multi-turn refinement using visual feedback from prior attempts achieves higher click precision and task success in GUI grounding for dense coding interfaces than single-shot prediction.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"12c4fcb6b2409764b3791491afea96a9bae99429f01d47a96fa568aacd7e933e"},"source":{"id":"2604.13019","kind":"arxiv","version":2},"verdict":{"id":"8cbffd65-8462-413f-965f-0b5aa4c6d479","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T15:54:16.329493Z","strongest_claim":"multi-turn refinement significantly outperforms state-of-the-art single-shot models in both click precision and overall task success rate.","one_line_summary":"Multi-turn visual feedback refinement outperforms single-shot coordinate prediction for pixel-precise GUI grounding in complex coding environments.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That models can reliably interpret and act on visual feedback from prior attempts to self-correct without introducing additional errors or failing to adapt to dynamic UI changes.","pith_extraction_headline":"Multi-turn refinement using visual feedback from prior attempts achieves higher click precision and task success in GUI grounding for dense coding interfaces than single-shot prediction."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.13019/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}