Paste the shared instructions once, followed by this prompt and the permitted source material.

Separate the educational subject, the work performed and the contribution established.
1. Identify the actual manuscript type and computing-education focus. Use the supplied current Manuscript Types page; do not invent an accepted category or add computing relevance the work does not have.
2. State the strongest contribution already supported, in plain language, and locate it in the supplied body. Do not mistake a tool name, method, topic or future use for a result. At planning stage, identify an intended contribution as intended, not accomplished.
3. Compare it with the nearest work supplied or actually verified. Specify the difference in understanding, evidence, judgement or available capability; otherwise mark novelty unverified. Do not infer originality from the title or lack of a matching citation.
4. Identify who could use this knowledge and for what particular purpose. Do not invent a classroom application.

Return: a short scope/type account; one supported contribution sentence with location; the comparison or verification limit; and up to three material issues. Distinguish wording repair from missing evidence, reasoning or contribution. A title-only assessment is provisional.

Input: [GUIDANCE + FRONT MATTER + RELEVANT BODY EXCERPTS + NEAREST-WORK MATERIAL, IF AVAILABLE]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Trace: problem -> aim/question/thesis -> approach -> evidence or argument -> interpretation -> contribution -> educational significance and implications.

Adapt this chain to the actual inquiry. A conceptual argument need not contain participants or statistical results; interpretation may be developed without a section named Discussion. Do not require every paper to demonstrate a mechanism, causal effect or classroom intervention.

For each material link, identify its location and status: clear; underexplained; not supported; or cannot verify. Distinguish an absent inference from existing reasoning that is merely hard to find. Consistent repetition of a claim is not support for it.

Return a compact table: Link | Location/status | Consequence | Smallest defensible action. Highlight at most three substantive breaks. Do not manufacture the missing argument or prescribe additional analyses without identifying the retained claim that needs them.

Input: [AIM/THESIS + APPROACH + EVIDENCE/ARGUMENT + DISCUSSION + CONCLUSION]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Check the central claims, normally one to three, rather than every sentence.

For each, identify what was studied or argued, what material supports it, the reasoning connecting them and the essential qualification. Distinguish observed findings, conceptual conclusions, constructed examples, specified designs, tested capabilities, proposed uses and future benefits.

Self-report need not measure performance; assisted performance need not demonstrate independent learning; agreement or passing technical tests need not establish validity for an educational use. Different evidence types can complement one another without measuring the same thing. A nonsignificant result does not establish equivalence, and a significant result beside a nonsignificant one does not establish different effects.

Return: Claim | Source and reasoning | Supported / underexplained / overstated / not established / cannot verify | Minimum correction. Where narrowing is justified, suggest only a local alternative and say what evidence or interpretation survives. Adding “may” does not supply missing support. Do not infer a statistical or causal claim that the authors explicitly do not make.

Input: [CENTRAL CLAIMS + SUPPORTING MATERIAL + ESSENTIAL LIMITATIONS]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Identify the intellectual work of the Discussion wherever it occurs.

Does it explain what the findings or argument mean, relate that addition to the nearest scholarship, consider material alternatives, and show why it matters for computing education? For each main implication, locate its supporting reasoning and the people, question or judgement it concerns. Distinguish a proposed application from evidence that it worked.

Do not generate a theory, teaching recipe or new “contribution” paragraph that the study has not developed. A rigorous negative finding, conceptual distinction or contextual interpretation can matter without immediate implementation. Conversely, correcting the authors' own implementation does not automatically establish a general methodological advance.

Return: one plain-language statement of the supported insight and its location; up to three material gaps; and the smallest structural or claim correction for each. Suggest an outline only when organisation is the actual problem. Keep optional future research separate from changes needed for the current claims.

Input: [AIMS + MAIN FINDINGS/ARGUMENT + DISCUSSION + RELEVANT PRIOR WORK]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Audit first. Do not rewrite automatically.

1. Meaning and support. Compare title, complete abstract, keywords and contribution bullets with the supplied body. Locate the strongest contribution already established. Separate clear writing, support and independent scholarly value. Flag activity lists presented as results, intended benefits presented as achievements, and a limitation elsewhere that fails to qualify an overstated claim. Do not invent support or treat missing excerpts as absent scholarship.

2. Different jobs. The abstract must stand alone as a research summary, including the principal contribution, educational significance and warranted implication. Contribution bullets explain the meaning for computing teachers, lecturers, curriculum designers and interested readers outside the methodological speciality. Ask what they learn, why it matters to them, and what they can reasonably understand, judge, consider or investigate. Conceptual overlap is necessary; repeated study summaries, methods lists and promotional promises are not.

3. Reader test. Can that reader explain the particular result and its limits without decoding specialist terminology? Flag only obstructive jargon, repetition or generic benefit language. Preserve precision and indispensable terms. A conceptual insight, resource or supported warning can be useful without a classroom-effectiveness claim. Simpler wording must not make the conclusion stronger.

4. Format, using the applicable supplied policy:
- Title: informative and consistent with the paper; approximately 12–20 words is guidance, not a cutoff.
- Abstract: 150–250 words excluding its heading and labels; four short paragraphs headed Problem and aim; Approach; Findings; Conclusions. Main argument may replace Findings for conceptual/theoretical papers; Synthesis may replace it for reviews. Use one third label. No citations or unexplained abbreviations.
- Keywords: 4–6 phrases; each phrase counts once. Natural overlap with the title is acceptable.
- Contribution and implications for informatics education: immediately after the keywords; 3–5 full-sentence bullets, at most 150 words including labels, excluding the section heading. Introduce Knowledge contribution; Educational and community value; Implications for CS education in that order, each at least once. Extra bullets may repeat a relevant label; Conditions and limits is optional as a final bullet. No compulsory fourth bullet or minimum word count. Check bold labels only when formatting is visible. Do not include citations.

Apply the 8 October 2026 receipt-date transition in the supplied policy. Older manuscripts receive only the editor's specified presentation updates; do not reopen settled science. Counts and labels do not establish contribution quality.

Return: the separate counts and counting uncertainty; a compact table of format, reader clarity and support; and up to three consequential issues. Distinguish reporting repair, unsupported claims and unavailable evidence. Keep source locations in the audit, not in publication-ready bullets.

Input: [POLICY/RECEIPT DATE + COMPLETE FRONT MATTER + SUPPORTING BODY PASSAGES]

Use only after the support check. This is language assistance, not permission for AI to create findings or educational value.

Use this only after the authors have checked the facts, interpretation and support. Edit the supplied verified notes; do not discover or manufacture a contribution.

Prepare only “Contribution and implications for informatics education” for computing educators and interested non-specialists, not a second technical abstract. Follow the supplied format: 3–5 full-sentence bullets, at most 150 words including labels. Introduce Knowledge contribution; Educational and community value; Implications for CS education in that order. Extra bullets are optional; Conditions and limits is not compulsory.

State the specific finding or insight, who needs it and why, and the implication it supports. Retain essential context, scope, uncertainty and the distinction between results and proposed uses. Do not add mechanisms, classroom applications, comparisons, numbers, novelty claims or benefits that the verified notes do not establish. Do not turn a feature list into a finding or conceal failed validation behind “promising”. Avoid citations, unexplained abbreviations and generic claims about improving education.

If the notes do not support a contribution, explain the precise gap instead of filling it with prose. Missing source access is a verification limit, not proof of absence. If sufficient, return only the draft bullets, their word count, and a brief separate source map for author checking. Do not append the source map to the manuscript's contribution section.

Input: [AUTHOR-VERIFIED FINDING/INSIGHT + VALUE + IMPLICATION + QUALIFICATIONS + SUPPORTING PASSAGES + APPLICABLE FORMAT]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Check the reporting appropriate to the actual study: approval, exemption, waiver or no-review-required basis; consent and voluntariness; privacy; data/material access; funding/conflicts; AI use; and blind-review placement.

Distinguish consent from ethics review and permission to reuse records. Written, oral or electronic consent must match the actual procedure and applicable rules; adulthood does not itself establish exemption. A later approval date does not alone prove misconduct: distinguish original collection, subsequent research use and the scope of the decision. Flag a specific unresolved inconsistency rather than demand a new approval or retrospective consent.

Separate a statement missing from the excerpt from an actual missing basis. Clear, consistent author reporting is not independent authentication, but do not routinely request certificates or signed forms. Where the basis is genuinely uncertain, identify the precise question for the responsible institution. Diagnose reporting, not legal compliance.

Describe consequential AI use accurately: language support, coding/analysis, translation, generated material and participant-data processing are different. Do not invent a disclosure or assume anonymisation from removal of names.

Return only applicable issues: Item | Location/verification limit | Why it matters | Minimum clarification. Keep confidential identifying information out of the review manuscript and public AI tools.

Input: [REDACTED METHODS + EXISTING DECLARATIONS + APPLICABLE RULES, IF AVAILABLE]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Compare the manuscript's evidence configuration with the supplied INFEDU Casebook entries, not just their titles.

Select at most three relevant cases. For each, identify the matched case/tag, the particular evidence or inference issue, and the smallest applicable reporting or claim correction. A match is diagnostic guidance, not an editorial decision or a compulsory study design. Do not require a pilot, extra coder or larger sample solely because a case mentions one.

Do not reconstruct Casebook rules from a tag, a search snippet or memory. If the entries are unavailable, say “Casebook comparison cannot be verified” and identify which entry needs reading. If none fits, say “no strong Casebook match”. Do not force a match or invent a new official case.

Return: Case | Why it fits | Supplied passage | Relevant action or qualification. Keep already-resolved matters out of the action list.

Input: [MANUSCRIPT SUMMARY + CLAIM + ACTUAL CASEBOOK EXCERPTS]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Separate the empirical and measurement claims and identify which is primary. For each, trace the construct, data, scoring, analysis and inference actually supported.

Check the instrument's source, adaptations and translations; the meaning of total/subscale scores; relevant evidence for their intended interpretation and use; and any overlap between predictor and outcome content. Reliability alone is not full validation, but do not demand every possible validation study or latent-factor model when the retained claim does not need it.

Identify whether weak measurement undermines the empirical conclusion, or whether a narrower conclusion remains supported. Keep reported attitudes, assessed performance and learning distinct. Note available instruments and restricted access accurately; do not assume unprovided files are absent.

Return: Claim/component | Located support | Consequence of the gap | Minimum repair or defensible narrowing. Do not request a second full study by default.

Input: [AIMS + METHODS + INSTRUMENT/SCORING + RESULTS + CONCLUSIONS]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Identify which sources directly address the computing-education question and which supply comparison, theory or adjacent evidence. Do not treat source labels or study counts as a synthesis.

Check whether selection, extraction and interpretation are traceable to the degree appropriate to the review design. Does the synthesis develop an insight, distinction, explanation or corrective conclusion beyond a list of topics? Locate the reasoning and distinguish study findings from the review authors' interpretation.

Check whether the title, abstract and recommendations imply stronger or more direct evidence than the corpus supplies. Compare the addition with nearest reviews only where their relevant content is supplied or actually verified; otherwise mark that comparison unverified. Do not fabricate a search or require meta-analysis for a conceptual or qualitative synthesis.

Return a short account of the supported synthesis and up to three material issues, with locations and minimum corrections. An evidence gap can be a worthwhile finding when its basis and consequence are established.

Input: [QUESTION + METHOD + CORPUS/EVIDENCE TABLE + SYNTHESIS + CONCLUSIONS]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Identify who or what was studied and what the AI actually did. Distinguish generated-output quality, assisted task performance, learner reasoning, independent learning and perceptions.

Check the reported model/tool version, dates, prompts/settings and relevant outputs only to the extent needed for the claims. For model-only studies, do not invent learner participation. For mixed evidence, check how the sources are connected rather than assume that parallel summaries confirm each other.

Examine any central scoring rule or reference label: an automatic check, a passing test suite or agreement with another model does not by itself establish correct educational feedback. A failed check can be informative, but identify the supported lesson beyond repairing one implementation.

Flag causal or general-use claims only when actually made. A comparator, new model or classroom experiment is not mandatory for every exploratory/design contribution. Narrowing must leave a genuine result, not a generic promise.

Return up to three issues with locations, their effect on the conclusion and the smallest defensible action; identify remaining verification limits.

Input: [DESIGN + AI SETUP + TASKS + EVIDENCE + CONCLUSION]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Identify whether the work establishes a conceptual argument, an explicit specification, a demonstrated capability, an expert appraisal or an observed outcome. These are different contributions; none automatically proves learning effectiveness.

Trace the important design or evaluation claim to the actual criteria, material, reviewer roles, procedure and finding. Does the paper explain the decision or interpretation beyond naming a new framework? Inspect worked examples for what they genuinely establish. Agreement and validity are distinct; do not require a reliability coefficient or extra reviewers unless the claim depends on reproducible classification.

Check whether feedback demonstrably informed the design when that is claimed, without inventing iteration history. A completed, justified specification need not acquire a classroom trial. An error found in its own implementation is not automatically a new general principle.

Return: Claim | Evidence/argument | What follows | Minimum clarification. End with one supported contribution sentence, or state precisely what cannot be established from the supplied material.

Input: [PURPOSE + FRAMEWORK/PROTOCOL + EVALUATION RECORD + CONCLUSION]

Paste the shared instructions once, followed by this prompt and the permitted source material.

Check the current version against the stated revision actions and supplied policy. Focus on unresolved issues and direct consequences of changes; do not generate a fresh wish list. Keep resolved matters closed unless the current text supplies a specific material reason to revisit them.

Check consistency and support of the central claim across front matter and body, reader-facing contribution language, and consequential reporting/access issues. Use the supplied current abstract/keyword/bullet rules or the completed P05 audit; do not treat correct labels as proof of scholarly quality. Do not endorse novelty from a summary or from AI reassurance.

Apply the receipt-date/transition rule and any editor-agreed length exception. Record the supplied body word count as information. INFEDU's 6,000–8,000-word target and general 10,000-word guidance concern the final version, not an initial-submission eligibility cutoff. Flag repetition only where visible; do not require numerical shortening by default.

Return up to three unresolved material issues, each with location and minimum action; list essential checks that could not be performed separately. End with one summary: no major issue identified in the material checked; specific issues remain; or insufficient material for a meaningful final check. None is an acceptance prediction, submission clearance or substitute for human review.

Input: [CURRENT EXCERPTS + CHANGE LOG + POLICY + EXISTING DECISION, WHERE APPLICABLE]