30 Claude Prompts for a Research Proposal You Can Defend

<context> I need to pick a study design I can actually run. Do not invent a methods paper or a published example I did not paste. </context> <inputs> - Research question: [RESEARCH QUESTION] - Field: [FIELD] - Data I can reach: [DATA ACCESS] - Months I have: [MONTHS] - Skills I have: [SKILLS I HAVE] - Designs I am considering: [CANDIDATE DESIGNS OR UNDECIDED] </inputs> <task> Recommend one design I can run. For each candidate (or for 4 common families if I said UNDECIDED): typical aim, when it fits, workload, and a reason to reject it. End with one recommendation, what it can support, what it cannot, and the first three protocol tasks. </task> <constraints> - Do not invent papers, authors, years, or a precedent study. - Do not invent access to a dataset, instrument, or lab I did not name. - Be honest if [MONTHS] cannot support a full trial, ethnography, or multi-wave survey. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Comparison table, then a 8-line recommendation, then UNVERIFIED rows. </format>

<context> I need a design justification a committee or funder can live with. Argue from the question and the constraints I name. Do not invent a methods paper I did not paste. </context> <inputs> - Research question: [RESEARCH QUESTION] - Design: [STUDY DESIGN] - Alternatives I considered: [ALTERNATIVES] - Constraints: [DATA ACCESS], [MONTHS] - Notes: [PASTE NOTES] - Methods source I already have (optional): [PASTE METHODS SOURCE OR NONE] </inputs> <task> Justify [STUDY DESIGN] for [RESEARCH QUESTION]: why it fits, what it can support, what it cannot, alternatives I rejected and why, and the threats I already accept. If I pasted a methods source, you may cite only that label. </task> <constraints> - Do not invent papers, authors, years, DOIs, or quotations. - Do not pretend a qualitative design can produce a causal estimate I did not measure. - If the design cannot answer the question, say so and propose a smaller question, not a fake method. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Justification memo, then a cannot-support list, then a one-paragraph "I chose this because" draft, then UNVERIFIED rows. </format>

<context> I will paste access notes and, if I have it, a planned or known headcount. Describe only that sample or corpus. Do not complete a missing n from a typical paper in my field. </context> <inputs> - Research question: [RESEARCH QUESTION] - Planned or known n: [SAMPLE N OR UNKNOWN] - Access notes: [PASTE ACCESS NOTES] - Inclusion rule: [INCLUSION RULE] - Setting: [SETTING] - Who I cannot reach: [MISSING POPULATION] </inputs> <task> Write a sample or corpus subsection: who or what is in the set, how they enter, who is missing, and the n I supplied. If [SAMPLE N OR UNKNOWN] is UNKNOWN, keep every sentence n-free and list the cells I must open or negotiate. </task> <constraints> - Do not invent n, a response rate, a demographic split, or an archive I did not name. - If a cell is missing, write NOT IN PASTE. - Do not upgrade a convenience sample into a representative one. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Sample or corpus prose, then a table of known vs NOT IN PASTE fields, then UNVERIFIED rows. </format>

<context> I need an analysis plan for a proposal, not results. Do not invent a test, a model, or an n I did not name. </context> <inputs> - Research question: [RESEARCH QUESTION] - Design: [STUDY DESIGN] - Variables, codes, or materials I already named: [PASTE ANALYSIS NOTES] - Software I can use: [SOFTWARE] - Sample n if known: [SAMPLE N OR UNKNOWN] </inputs> <task> Write an analysis plan: units, preparation steps, primary analysis, secondary or exploratory work, how I will handle missingness if I said anything about it, and what I will not claim. Leave every count as UNKNOWN unless I pasted it. </task> <constraints> - Do not invent n, p-values, effect sizes, or a finding the study has not produced. - Do not add a test I cannot run in [SOFTWARE] unless you label it OPTIONAL, NEEDS NEW TOOL. - If [SAMPLE N OR UNKNOWN] is UNKNOWN, refuse a power claim. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Numbered analysis steps, then a will-not-claim list, then UNKNOWN cells, then UNVERIFIED rows. </format>

<context> I need ethics and risk notes that match what I am allowed to do. Do not invent an approval number, a waived review, or a population I cannot reach. </context> <inputs> - Research question: [RESEARCH QUESTION] - Ethics status: [ETHICS ID OR NONE / PENDING / NOT REQUIRED] - Known risks: [KNOWN RISKS] - Population I could not reach: [MISSING POPULATION] - Data type: [HUMAN / ARCHIVAL / PUBLIC / OTHER] </inputs> <task> Draft ethics and risk notes: approval status only as I stated it, consent or data-use notes from the paste, risks I will manage, and limits I already accept. Separate ethics facts from study limits. </task> <constraints> - Do not invent an IRB, REB, or ethics ID. - If [ETHICS ID OR NONE / PENDING / NOT REQUIRED] is NONE, write that the proposal must not claim approval. - Do not invent a risk-closing procedure I did not name. - Include the limit of using Claude on this proposal. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Two subsections (ethics / risks), then 8 sentences I can reuse if they survive checking, then UNVERIFIED rows. </format>

<context> I need a proposal timeline that fits the date I actually have, not an optimistic novel. Do not invent a committee meeting or a data-release date I have not booked. </context> <inputs> - Venue: [THESIS COMMITTEE / GRANT / COURSE] - Hard deadline: [DEADLINE] - Current stage: [CURRENT STAGE] - Hours I can work per week: [HOURS PER WEEK] - Dates I have already booked: [BOOKED DATES OR NONE] - Ethics if needed: [ETHICS STATUS] </inputs> <task> Build a milestone timeline from [DEADLINE] and [CURRENT STAGE]: question lock, ethics if needed, access, data or fieldwork, analysis, each proposal section, reviewer rounds, and a buffer. Mark any date I did not supply as UNBOOKED. </task> <constraints> - Do not invent a booked slot, a data drop, or n weeks already completed. - If [HOURS PER WEEK] cannot hit [DEADLINE], say so and propose cuts, not a 70-hour week. - Buffer must be visible, not hidden inside each task. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Dated list with UNBOOKED tags, then a one-page risk note, then UNVERIFIED rows. </format>

<context> I need a hostile feasibility check before I promise the study. Do not invent access, a skill, or a dataset that would make the larger version easy. </context> <inputs> - Research question: [RESEARCH QUESTION] - Design: [STUDY DESIGN] - Months I have: [MONTHS] - Data I can reach: [DATA ACCESS] - Skills I have: [SKILLS I HAVE] - Known blockers: [KNOWN BLOCKERS OR NONE] </inputs> <task> Stress-test feasibility across time, access, skills, ethics, and reviewer bandwidth. For each risk: likelihood from the paste only, what it blocks, a cut that would save the proposal, and a stop rule if the blocker is fatal. End with a go / shrink / stop verdict. </task> <constraints> - Do not invent access to a dataset, archive, lab, or translator I did not name. - Be blunt if [MONTHS] cannot support the current question. - Do not invent papers that would prove the design is standard. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Risk table, then go / shrink / stop, then a cut list, then UNVERIFIED rows. </format>

<context> I need a risk map for access and skills, not a pep talk. Use only what I paste. Do not invent a portal, a license, or a training I did not name. </context> <inputs> - Research question: [RESEARCH QUESTION] - Data I can reach: [DATA ACCESS] - Skills I have: [SKILLS I HAVE] - Skills I still need: [SKILLS I NEED] - Language or software limits: [TOOL LIMITS] </inputs> <task> Map access and skill risks: data or site access, language, software, training time, and backup plans that use only tools I named. For each risk: what fails, an early test I can run this week, and whether the proposal should disclose it. </task> <constraints> - Do not invent a dataset, vendor, or course that would close the gap. - If you are unsure a tool exists, write VERIFY ACCESS, do not name a fake portal. - Do not invent n days a skill "usually" takes unless I pasted that. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Risk cards, then a this-week test list, then UNVERIFIED rows. </format>

<context> I need a send-and-wait plan for real people who will read this proposal. Do not invent that someone has already agreed to a date I did not book. </context> <inputs> - Readers: [COMMITTEE OR ADVISOR LIST] - Deadline: [DEADLINE] - What is ready to send: [READY SECTIONS] - Advisor preference on order: [ADVISOR NOTES OR NONE] - Typical wait in my program: [WAIT DAYS OR UNKNOWN] </inputs> <task> Sequence feedback rounds for [COMMITTEE OR ADVISOR LIST] against [DEADLINE]: what I send each person, in what order, how long I should wait, and what I do if a reader is silent. Mark every date I did not confirm as UNBOOKED. </task> <constraints> - Do not invent a member's availability or a publication they wrote. - Do not invent n days they usually take unless I pasted that. - If [WAIT DAYS OR UNKNOWN] is UNKNOWN, use a blank I must fill from my program, not a guessed average. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Ordered send plan with UNBOOKED dates, then a silence protocol, then UNVERIFIED rows. </format>

<context> The current question does not fit the calendar. I need specific cuts, not a vague smaller project. Do not invent a dataset that would save the larger version. </context> <inputs> - Research question: [RESEARCH QUESTION] - Months I have: [MONTHS] - Deadline: [DEADLINE] - Tempting extras: [SIDE TOPICS] - Data I can reach: [DATA ACCESS] - What I refuse to drop: [MUST KEEP] </inputs> <task> Cut the project to fit [MONTHS] and [DEADLINE]. Give: (1) a smaller question, (2) a drop list with a reason for each cut, (3) what [MUST KEEP] still costs in time, (4) a one-sentence scope I can put in the proposal, (5) a warning if [MUST KEEP] still blows the calendar. </task> <constraints> - Do not invent access or a method that would keep the large version alive. - The drop list should be specific, not "everything else." - Do not invent papers that justify the smaller question. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Smaller question, drop table, cost of [MUST KEEP], scope sentence, UNVERIFIED rows. </format>

<context> I need likely reviewer or committee objections from the proposal I actually wrote, not from a generic oral-exam list. Do not invent a finding they will ask me to defend. </context> <inputs> - Research question: [RESEARCH QUESTION] - Proposal notes or draft: [PASTE PROPOSAL NOTES] - Venue: [THESIS COMMITTEE / GRANT / COURSE] - Known reviewer interests: [REVIEWER NOTES OR NONE] - Weakest section as I see it: [WEAK SECTION] </inputs> <task> From the paste, draft 10 objections: gap, design fit, access, n or corpus, analysis, ethics, timeline, and contribution. For each: why it is likely, a 4-sentence honest reply, and what I must still verify. No bluff numbers. </task> <constraints> - Do not invent a reviewer's publication, a paper, or a result I must remember. - Do not invent n or a citation to fill silence. - If the paste is too thin to answer, write NEED NOTES, not a guess. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Numbered objections with why / draft reply / verify list / UNVERIFIED tags. </format>

<context> Reviewers reject proposals when the gap is louder than the notes. Stress-test my gap. Do not invent a missing classic that would make it look earned. </context> <inputs> - Research question: [RESEARCH QUESTION] - Gap I think I can defend: [GAP STATEMENT] - Literature I have actually read: [PASTE NOTES] - Searches I already ran: [SEARCH NOTES OR NONE] </inputs> <task> Audit [GAP STATEMENT] against [PASTE NOTES]. Tag the gap STRONG / STRETCH / NOT IN PASTE. If it is stronger than the notes, rewrite a weaker gap I can defend. List searches I still owe as types, not as invented titles. </task> <constraints> - Do not invent papers, authors, years, DOIs, or quotations. - Do not claim "nobody has studied X" unless the paste supports "not in this set." - Do not fill a hole with a recalled famous paper. Write SEARCH FOR that topic instead. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Verdict, weaker gap rewrite, search-to-run list, UNVERIFIED rows. </format>

<context> I need a hostile methods audit: does the plan claim more than the design and access can support? Do not repair an overclaim with extra data I did not collect. </context> <inputs> - Research question: [RESEARCH QUESTION] - Methods draft or notes: [PASTE METHODS] - Data I can reach: [DATA ACCESS] - Design: [STUDY DESIGN] - Sample n if known: [SAMPLE N OR UNKNOWN] </inputs> <task> Audit [PASTE METHODS] for overclaims: causal language, dual coding I did not name, a representative sample I did not have, a test I cannot run, or a completed result written in past tense. For each: quote the line, name the overclaim, and rewrite a weaker line I can keep. </task> <constraints> - Do not invent n, a second coder, an instrument, or a finding. - Do not invent a methods citation to bless the design. - If [SAMPLE N OR UNKNOWN] is UNKNOWN, refuse a power or representativeness claim. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Table: quoted line | overclaim | weaker rewrite | UNVERIFIED? </format>

<context> I need to see which contribution claims I can actually show. Do not invent a result that would make a weak claim look useful. </context> <inputs> - Research question: [RESEARCH QUESTION] - Contribution claims I want to make: [PASTE NOTES] - Design: [STUDY DESIGN] - Data I can reach: [DATA ACCESS] - Literature I have: [PASTE SOURCE LIST OR NONE] </inputs> <task> Map each contribution claim as SUPPORTED (the design and paste can carry it), STRETCH, or NOT SUPPORTED. For STRETCH and NOT SUPPORTED, write a smaller claim I can keep or say DROP. Do not add a paper that would save a claim. </task> <constraints> - Do not invent papers, data, or a finding the study has not produced. - Do not treat expected outcomes as completed results. - Cite only labels that appear in [PASTE SOURCE LIST OR NONE]. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Table: claim | tag | smaller claim or DROP | UNVERIFIED. Then a clean contribution paragraph using SUPPORTED lines only. </format>

<context> I need a limitations subsection that matches the study I can run. A gap is something this proposal will not do. Do not invent the paper that would close it. </context> <inputs> - Research question: [RESEARCH QUESTION] - Known limits: [KNOWN LIMITS] - Data I can reach: [DATA ACCESS] - Sample n if known: [SAMPLE N OR UNKNOWN] - What I still wish I had: [WISH LIST] </inputs> <task> Write limitations from [KNOWN LIMITS] and [DATA ACCESS]: sample, measure, design, access, time, and my own analysis choices. For each limit: what it blocks, what it does not block, and whether I already flagged it in methods. No new citations. </task> <constraints> - Do not invent n. If [SAMPLE N OR UNKNOWN] is UNKNOWN, say the count is not in this paste. - Do not invent a missing classic or a next study with a fake design n. - Prefer "this set cannot speak to X" over "the field has never." - Include the limit of drafting with Claude. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Numbered limits with a blocks / does-not-block pair, then 6 reusable sentences, then UNVERIFIED rows. </format>

<context> I need a problem statement that funnels to my question. Do not open with an invented statistic or a famous paper I did not paste. </context> <inputs> - Research question: [RESEARCH QUESTION] - Scope statement: [SCOPE STATEMENT] - Why it matters, from my notes: [PASTE NOTES] - Gap I can defend: [GAP STATEMENT] - Citation style: [CITATION STYLE] - Word count: [WORD COUNT] </inputs> <task> Draft a problem statement of about [WORD COUNT]: field context only from my notes, why the topic matters only with support I pasted, the gap I can defend, and a last sentence that states [RESEARCH QUESTION]. No hanging claims. </task> <constraints> - Do not invent papers, authors, years, DOIs, quotations, or prevalence numbers. - If a sentence needs a source I did not paste, mark it UNVERIFIED or drop it. - Last sentence must be the research question, not a teaser. - Do not survey a literature I have not pasted. </constraints> <format> Continuous prose, then a claim-to-source checklist with UNVERIFIED rows. </format>

<context> I need a short literature section for a proposal, not a full review chapter. Cite only sources I list. Do not invent page numbers or quotations. </context> <inputs> - Research question: [RESEARCH QUESTION] - Sources I will cite: [PASTE SOURCE LIST] - Findings I verified: [PASTE FINDINGS] - Gap I can defend: [GAP STATEMENT] - Citation style: [CITATION STYLE] - Word count: [WORD COUNT] </inputs> <task> Draft a literature section of about [WORD COUNT] that synthesizes the listed sources around claims, not paper-by-paper summaries. Close on [GAP STATEMENT] only if the paste can carry it. If a section has fewer than two listed sources, mark it THIN. </task> <constraints> - Do not invent papers, authors, years, DOIs, quotations, or findings. - If a finding is not in [PASTE FINDINGS], omit it. - Use [CITATION STYLE] in-text forms only from the details I provided. - If a claim needs a source I did not give, mark it UNVERIFIED. - Do not claim "nobody has studied X" unless the paste supports "not in this set." </constraints> <format> Prose section, then a list of every in-text cite for me to check, then THIN and UNVERIFIED rows. </format>

<context> I will paste the protocol I can run, not results. The methods narrative must match the work I can do. Do not invent a second coder, a larger n, or a test I will not run. </context> <inputs> - Research question: [RESEARCH QUESTION] - Protocol I can run: [PROTOCOL] - Sample n if known: [SAMPLE N OR UNKNOWN] - Tools I will use: [TOOLS] - Citation style: [CITATION STYLE] - Word limit: [WORD COUNT] </inputs> <task> Write a methods narrative from [PROTOCOL]: design, participants or corpus, collection, analysis, validity or trustworthiness, ethics, and limits. Leave counts as UNKNOWN unless I pasted them. Use future or planned verbs, not fake past-tense results. </task> <constraints> - Do not invent n, instruments, software, or a second reviewer. - If I will work alone, say so. Do not upgrade that to dual independent coding. - If a count is missing, write UNKNOWN, not a rounded guess. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Methods draft under [WORD COUNT], then UNKNOWN cells I must fill, then UNVERIFIED rows. </format>

<context> I need the timeline and resources section of the proposal. Use only dates, people, and tools I named. Do not invent a budget line or a staff role I did not list. </context> <inputs> - Research question: [RESEARCH QUESTION] - Hard deadline: [DEADLINE] - Current stage: [CURRENT STAGE] - Data I can reach: [DATA ACCESS] - Resources I have: [RESOURCES I HAVE] - Budget notes if any: [BUDGET NOTES OR NONE] </inputs> <task> Draft a timeline and resources section: milestones with UNBOOKED tags where I gave no date, people and tools from [RESOURCES I HAVE], and budget lines only from [BUDGET NOTES OR NONE]. Flag any milestone that needs a resource I did not name. </task> <constraints> - Do not invent a grant amount, a salary, a software license, or a collaborator. - Do not invent a booked date. - If [BUDGET NOTES OR NONE] is NONE, write NEED BUDGET NOTES instead of a sample budget. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Timeline table, resources list, NEED BUDGET NOTES or budget lines from the paste, UNVERIFIED rows. </format>

<context> I need expected outcomes for a proposal, not fake results. Outcomes are types of findings I can produce. Do not write them as completed data. </context> <inputs> - Research question: [RESEARCH QUESTION] - Aims: [AIMS] - Design: [STUDY DESIGN] - What success looks like to me: [SUCCESS NOTES] - Word count: [WORD COUNT] </inputs> <task> Draft expected outcomes of about [WORD COUNT]: what kind of evidence each aim can produce, what I will deliver (report, dataset, codebook, chapter), and what I will not claim. Use future tense. No invented n, percent, or quotation. </task> <constraints> - Do not invent results, tables, or a finding I have not run. - Do not write past-tense results for a study that has not started. - If [SUCCESS NOTES] overclaims the design, shrink the outcome and say why. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Expected-outcomes prose, then a will-not-claim list, then UNVERIFIED rows. </format>

<context> I need a proposal outline a committee or funder can approve, not a finished proposal. Do not stock the outline with papers or results I have not pasted. </context> <inputs> - Research question: [RESEARCH QUESTION] - Venue: [THESIS COMMITTEE / GRANT / COURSE] - Notes I already have: [PASTE NOTES] - Word count target: [WORD COUNT] - Required headings if I have them: [VENUE HEADINGS OR NONE] - Citation style: [CITATION STYLE] - Deadline: [DEADLINE] </inputs> <task> Build an outline: problem, question, short literature job, aims, methods, timeline, ethics, expected outcomes, and what I still owe. Each section needs a claim or a job, a word budget, and a blank where a citation or count is missing. If I pasted [VENUE HEADINGS OR NONE], use those headings. </task> <constraints> - Do not invent citations, n, or a result the study has not produced. - If a section has no pasted support, leave it as a heading plus NEED NOTES. - Word budgets must sum to [WORD COUNT]. - If a claim needs a source I did not give, mark it UNVERIFIED. - Do not invent a funder heading I did not paste. </constraints> <format> Nested outline with word counts and NEED NOTES tags. No prose padding. UNVERIFIED list at the end. </format>

<context> I will paste the notes I actually have. Draft a complete research proposal from that paste only. Do not invent papers, data, or a literature I have not supplied. </context> <inputs> - Research question: [RESEARCH QUESTION] - Venue: [THESIS COMMITTEE / GRANT / COURSE] - Notes, sources, and protocol I have: [PASTE NOTES] - Word count: [WORD COUNT] - Citation style: [CITATION STYLE] - Required headings: [VENUE HEADINGS OR NONE] </inputs> <task> Draft the full proposal under [WORD COUNT]: problem, gap, aims, short literature, methods, timeline, ethics, expected outcomes, and limits. Cite only sources present in the paste. Tag any sentence that still needs a source I must find as UNVERIFIED. Leave counts as UNKNOWN unless I pasted them. </task> <constraints> - Do not invent papers, authors, years, DOIs, quotations, or data. - Do not write completed results for a study that has not run. - If a heading has no pasted support, write NEED NOTES under it instead of filling it. - Prefer a thinner honest draft over a complete-looking proposal I cannot evidence. </constraints> <format> Full proposal under the headings I gave (or a standard set if I said NONE), then a master UNVERIFIED and UNKNOWN list. </format>

<context> I must cut [PASTE DRAFT] to [WORD COUNT] or [PAGE LIMIT] without losing the argument or inventing tighter evidence. </context> <inputs> - Draft: [PASTE DRAFT] - Target word count: [WORD COUNT] - Page limit if any: [PAGE LIMIT OR NONE] - Must keep: [MUST KEEP] - Citation style: [CITATION STYLE] </inputs> <task> Cut to [WORD COUNT] (and [PAGE LIMIT OR NONE] if I set one). Protect [MUST KEEP]. Prefer deleting repetition, throat-clearing, and doubled examples. Do not delete a citation and keep the claim as if it were mine. Show a cut log. </task> <constraints> - Do not add new citations while cutting. - Do not merge two sources into one name. - Do not invent a tighter statistic to save space. - If the question cannot survive the cut, say so and stop before you gut the evidence. - If a claim needs a source I did not give, mark it UNVERIFIED. - Report old and new word counts. </constraints> <format> Cut draft, word counts, cut log (what left and why), UNVERIFIED rows. </format>

<context> I need a checklist before I submit a research proposal under [VENUE RULES]. This is a QA list, not a new draft of the proposal. </context> <inputs> - Research question: [RESEARCH QUESTION] - Venue: [THESIS COMMITTEE / GRANT / COURSE] - Venue rules: [VENUE RULES] - Word or page limit: [WORD COUNT] - Citation style: [CITATION STYLE] - Optional draft: [PASTE DRAFT OR NONE] </inputs> <task> Build a checklist: question match, gap honesty, methods fit, word count vs [WORD COUNT], citation completeness against sources I pasted, [CITATION STYLE] consistency, timeline with UNBOOKED dates flagged, ethics wording, expected outcomes in future tense, and any rule in [VENUE RULES]. Mark items I must verify by hand. </task> <constraints> - Do not invent a funder or committee rule I did not paste. - Do not invent that my citations are complete. - If I pasted a draft, flag only problems you can see. If I did not, keep items as blank checks. - No suggested extra papers. - If a claim needs a source I did not give, mark it UNVERIFIED. </constraints> <format> Checkbox list grouped by argument, evidence, style, venue. Hand-verify items in italics. UNVERIFIED rows last. </format>

<context> I need an abstract and a short cover note for [VENUE] based on a draft I actually have. No sales fiction and no invented papers in the journal or program. </context> <inputs> - Draft or near-full proposal: [PASTE DRAFT] - Venue: [VENUE] - Word limit for abstract: [WORD LIMIT] - Contribution I can defend: [CONTRIBUTION I CAN DEFEND] - Why this venue (my reasons): [WHY THIS VENUE] - Required statements: [ETHICS / CONFLICTS / NONE] </inputs> <task> Write: (1) an abstract within [WORD LIMIT] from [PASTE DRAFT] only, (2) a short cover note that states what the proposal is, [CONTRIBUTION I CAN DEFEND], and why this venue using only reasons I supply. If the draft lacks methods, a timeline, or a limit, write NOT IN DRAFT instead of guessing. </task> <constraints> - Do not invent papers, authors, years, findings, n, or an editor's name. - Do not make the abstract more positive than the draft. - Do not claim fit from papers I did not name. - If a claim needs a source I did not give, mark it UNVERIFIED. - Count words and stop at the limit. </constraints> <format> Abstract, word count, cover note, NOT IN DRAFT list, UNVERIFIED rows. </format>