Check the research chain
Start from the problem or gap, then trace objectives, method, data/evidence, results, and claims. If the chain breaks, fluent writing will not make the work ready.
Student Edition v1.3
A bilingual workflow for using workspace AI tools such as Codex, Claude Code, or Antigravity to review a thesis or manuscript before the next advisor or committee round.
A thesis review is not just language editing. It must test whether the whole research argument holds together: problem, objectives, method, evidence, results, references, scope, and limitations.
Start from the problem or gap, then trace objectives, method, data/evidence, results, and claims. If the chain breaks, fluent writing will not make the work ready.
Every important claim needs support from manuscript evidence, precise citations, datasets, baselines, metrics, raw results, or verified compliance sources.
Ask whether the work fits the degree level and target venue, whether key references or baselines are missing, and whether claims exceed the evidence.
Read this companion handout before using the prompt pack. It explains the five sharp thesis-review tests, Output/Outcome, prior attempts, unresolved need, alternatives, and responsible AI-assisted review.
Problem → prior attempts/evidence → unresolved need → objective/RQ → alternatives → method → result/evidence → output → outcome/by-product → conclusion
The recommended path for a full thesis review. Let the AI read files in a workspace, organize missing inputs, and create review notes without editing the source thesis first.
Use Gemini, ChatGPT, or Claude webapps only for short checks such as title/abstract, a few literature paragraphs, or one method/results section.