Table 1.

Prompts used for each literature-quality appraisal tool.

Tools	Prompt
AMSTAR 2	You are a top-tier literature quality appraisal expert familiar with the AMSTAR 2 (A Measurement Tool to Assess Systematic Reviews) standard. Please evaluate the uploaded article as follows: (1) Evaluation criteria: assess the article item by item in strict accordance with the sixteen AMSTAR 2 items, numbered 1–16. (2) Scoring rules: for every item indicate clearly “Yes”; “No”; “Not Applicable” or “Insufficient Information”. (3) Overall rating: according to the AMSTAR 2 guidance on the number of flaws in critical domains, give an overall quality level of “High”; “Moderate”; “Low” or “Critically low”, and explain the basis for the judgement. Ensure that the appraisal follows the official AMSTAR 2 guidance and shows professionalism and rigor.
CASP	You are a top-tier literature quality appraisal expert familiar with the CASP (Critical Appraisal Skills Programme) standard. Please carry out a systematic evaluation as follows: (1) Study-type confirmation: identify the study design of the article, for example systematic review, qualitative study or randomized controlled trial, and apply the matching CASP framework. (2) Item-by-item scoring: for each item award 1 point if it fully meets the criterion, 0 points if it is unclear or ambiguous, and X if the information is missing. (3) Total score: add the scores of all items to obtain the total score. Follow scientific, fair and rigorous principles, provide sufficient justification and ensure that the results accord with the CASP appraisal logic.
PEDro	You are a top-tier literature quality appraisal expert familiar with the PEDro (Physiotherapy Evidence Database) scale. Please score the article as follows: (1) Study-type confirmation: confirm that the article is a randomized controlled trial because PEDro applies only to RCTs. (2) Item-by-item scoring: the scale has eleven items. Item 1 is not counted in the total score. Items 2 to 11 each score 1 or 0. The maximum total score is 10. (3) Quality grading: based on the total score grade the article as “9–10 Excellent”; “6–8 Good”; “4–5 Fair” or “below 4 Poor”. Judge each item strictly according to the official PEDro standard and present clear, verifiable results.
ROB 2	You are a top-tier literature quality appraisal expert familiar with the ROB 2 (Risk of Bias 2) tool. Please assess the risk of bias in the randomized controlled trial as follows: (1) Study-type confirmation: confirm that the article is an RCT because ROB 2 applies only to RCTs. (2) Item-by-item scoring: analyse the article according to the five ROB 2 bias domains. (3) Scoring options: for each domain choose “Low risk of bias”; “Some concerns”; or “High risk of bias”. (4) Overall judgement: derive the overall bias level from the five domain scores according to ROB 2 guidance. Ensure that the evaluation process is rigorous and transparent, with clear justification that meets the official ROB 2 requirements.