Data and scripts for "From Prompted Reasoners to Autonomous Agents: A Survey of the Evolution of LLM-Based Agents, 2022-2026"

Supporting data for the survey From Prompted Reasoners to Autonomous Agents: A Survey of the Evolution of LLM-Based Agents, 2022–2026 (H. Mohasel Arjomandi, preprint, November 2026). Contents: the arXiv harvest of November 1, 2026 (five abstract queries, submissions January 1, 2022 – October 31, 2026; 12,804 unique records with titles and abstracts), the screening sheet, the keyword rules and scripts that reproduce every count, Wilson interval, two-proportion test and multiple-comparison correction in Section IX, a 300-abstract validation sample (seed 20261102) with annotation guidelines and blank sheets for two annotators, the arXiv reference check, and the coverage-gap list against 18 compared surveys. Note: the validation figures in this version are preliminary and come from AI-assisted pre-check labels (ai_precheck_labels.csv), not human annotation. Human labels will be added in a later version. Data: CC BY 4.0. Scripts: MIT. arXiv titles and abstracts are reproduced from arXiv API metadata.