Rules Before You Start
Not recommended for a full thesis review
If you have Codex, Claude Code, or Antigravity, use the Workspace AI guide. Webapps are suitable only for short sections.
Do not paste the whole thesis and ask "is it good?"
Start with the title, abstract, problem, and objectives. Then review literature, method/results, ethics, and writing hygiene in separate rounds.
Do not let AI invent evidence
Do not invent citations, DOI, venue rankings, metrics, results, IRB/legal requirements, or legal statements. If evidence is missing, the AI must say so.
Copyable Prompts
Use one prompt at a time. Replace the bracketed placeholders with text from the manuscript.
Prompt 1: Missing input check
I want you to review this thesis/manuscript draft as an academic reviewer. First task: - Read only the text I provide. - Identify what information is missing. - Classify each missing item as critical / useful / not needed in this round. - If information is already in the document, summarize it from the document instead of asking me to fill it again. - Ask follow-up questions only when the missing information is necessary for review. Rules: - Do not invent citations, DOI, venue rankings, metrics, results, IRB/legal requirements, or legal statements. - If evidence is insufficient, say so. - Cite manuscript locations when possible: section, paragraph, table, or figure. - Separate output from outcome, and check whether the draft traces problem -> prior attempts/evidence -> unresolved need -> objective/RQ -> method -> evidence -> output -> outcome. [Paste title page, abstract, introduction/problem statement, objectives/RQs, table of contents, and methodology summary if available.]
Prompt 2: Title, abstract, problem, objectives
Review only the title, abstract, problem statement, and objectives/RQs below. Output: 1. Whether title and abstract align. 2. Whether the problem has enough academic or practical weight. 3. Whether objectives/RQs come from the problem. 4. Claims that are stronger than the evidence. 5. Top 5 fixes before sending to an advisor. Do not rewrite everything. Give specific issues and reasons. [Paste title, abstract, problem/gap paragraphs, and objectives/RQs.]
Prompt 3: Literature review with limited quota
Review this literature review section as a field reviewer. Focus on: 1. Synthesis vs paper-by-paper summary. 2. Claims that need citations. 3. Citations that do not support the claim. 4. Reference role clusters. 5. Comparison matrix needed. 6. Direct prior work vs cross-domain support vs foundational support vs target-venue anchors. Respond concisely with tables and bullets. [Paste related work headings, 1-2 key paragraphs per heading, cited references/BibTeX, and any existing comparison table.]
Prompt 4: Method, data, baseline, results
Review the methodology and results below. Answer: 1. Does the method address the problem/objectives? 2. Are baselines sufficient? If not, what is missing? 3. Are metrics appropriate? 4. Do results answer the objectives/RQs? 5. Which result claims lack raw results, scripts, error analysis, or measurement support? 6. What are the top 5 blocking issues? Also check problem survival, genuine failure, mandatory gates, measurement ceiling, and no-new-problem guardrails. Do not invent experimental results. [Paste methodology, data collection, baselines, evaluation metrics, results tables/figures, and relevant appendix summaries.]
Prompt 5: Ethics / IRB / legal flags
Flag possible ethics, IRB, and legal/compliance issues from the text below. Rules: - Do not provide legal advice. - Do not conclude that IRB is unnecessary unless institutional evidence is provided. - State what should be checked with the advisor, IRB office, institution, venue, or qualified compliance source. Output: 1. Possible IRB/ethics approval or exemption issues. 2. Consent, anonymization, or de-identification needs. 3. Dataset license, copyright, Terms of Service, ownership, or confidentiality issues. 4. Statements the manuscript may need: ethics, data availability, consent, AI use, or limitations. [Paste participants/data collection/user study/data source/license/scraping/ethics statement sections.]