--- name: Journal Cover Letter slug: journal-cover-letter category: Writing description: Create, revise, and benchmark cover letters for academic journal submissions using verified manuscript facts. Routes documents by research type (empirical, synthesis, bibliometric) for targeted editorial persuasion. github: "https://github.com/hujizhou35-cmd/journal-cover-letter-tutorial/tree/main/skills/journal-cover-letter-skill" language: HTML stars: 28 forks: 1 install: "npx degit https://github.com/hujizhou35-cmd/journal-cover-letter-tutorial/tree/main/skills/journal-cover-letter-skill ~/.claude/skills/journal-cover-letter-skill" installs_to: ~/.claude/skills/journal-cover-letter-skill source_path: skills/journal-cover-letter-skill/SKILL.md collection_size: 2 category_size: 1012 collection_url: "https://dirskills.com/collections/hujizhou35-cmd/journal-cover-letter-skill" added: 2026-08-11T07:21:57.447Z last_synced: 2026-08-11T07:21:57.447Z canonical_url: "https://dirskills.com/skills/journal-cover-letter" --- # Journal Cover Letter Create, revise, and benchmark cover letters for academic journal submissions using verified manuscript facts. Routes documents by research type (empirical, synthesis, bibliometric) for targeted editorial persuasion. **Install:** ```bash npx degit https://github.com/hujizhou35-cmd/journal-cover-letter-tutorial/tree/main/skills/journal-cover-letter-skill ~/.claude/skills/journal-cover-letter-skill ``` ## README # Journal Cover Letter Skill v3.2 Turn verified manuscript facts into a clear editorial decision aid. Default to English unless the user requests another language. Prioritize biomedical and life-science submissions while remaining useful across disciplines. ## Core architecture: evidence anchor -> editorial meaning Every persuasive claim must contain both parts: - `empirical_anchor`: the concrete, manuscript-traceable observation selected for the editor; - `editorial_meaning`: what that observation changes in understanding, interpretation, coordination, practice, or the next research decision. For manuscripts with a distinctive empirical taxonomy, trend set, or named result pattern, also preserve an `authorial_empirical_fingerprint`: the minimum manuscript-native detail that lets an editor distinguish this paper from another paper on the same topic. A letter fails when it offers facts without meaning, meaning without a traceable factual anchor, or a polished abstraction that erases the manuscript's empirical fingerprint. This is the central v3.1 rule. ## Core rules - Treat the manuscript and author-confirmed materials as the factual source of truth. Never invent titles, results, registrations, declarations, author details, editor names, or journal requirements. - Separate facts into `verified`, `conflict`, `missing`, and `not_applicable`. Separate factual claims from interpretation. - Use the strongest wording the evidence supports. Accuracy should sharpen the pitch, not make it timid. - Separate the journal's official submission label from the manuscript's intellectual route. They may differ. - Treat previous letters as user-controlled. Reuse or analyze them only within explicit permission. - Treat an expert-authored letter as evidence of selection and editorial judgment, not as a gold standard or factual authority. - When testing a skill against a human benchmark, freeze a blind baseline before revealing the benchmark, compare editorial effects rather than wording, and modify transferable rules rather than patching a single draft. - Verify current journal information from official sources after the factual foundation is stable. - Use bounded revision loops. Reaching a loop limit is not success. - Use scripts for deterministic extraction, validation, auditing, and DOCX generation. Keep scientific meaning and editorial judgment in model reasoning. ## 1. Intake, routing, and fact sheet Inventory the manuscript, title page, supplements, correspondence, previous cover letters, and user instructions. Record each important item with its source. Set: ```text official_article_type: exact journal label | UNRESOLVED intellectual_route: ORIGINAL_RESEARCH | REVIEW_SYNTHESIS | BIBLIOMETRICS | OTHER_OR_UNRESOLVED submission_branch: INITIAL | REVISION | RESUBMISSION | TRANSFER | INVITED | OTHER fact_status: verified | conflict | missing | not_applicable previous_letter_permission: FORMAT_ONLY | FORMAT_AND_TONE | MAXIMUM_SUITABLE_WORDING | FACT_CHECK_ONLY | ANONYMOUS_EXPERT_BENCHMARK | NONE ``` ### Route by the primary evidence object - `ORIGINAL_RESEARCH`: the central evidence comes from participants, specimens, experiments, observations, surveys, clinical records, models, or other primary empirical data. - `REVIEW_SYNTHESIS`: the central evidence comes from the substantive findings of included studies, and the contribution is a changed interpretation, boundary, taxonomy, or evidence-based agenda. - `BIBLIOMETRICS`: the central evidence comes from bibliographic metadata, citations, terms, authorship, affiliations, journals, or networks, and the contribution is a map of field structure, evolution, concentration, connectivity, or emerging attention. Do not infer the intellectual route from the journal's article-type label. A bibliometric manuscript may be submitted under `Original Research`, `Review`, `Systematic Review`, or another journal-specific label. Record both fields explicitly. For hybrid manuscripts, choose the route that carries the main editorial contribution. If bibliometric analyses merely support a substantive evidence synthesis, use `REVIEW_SYNTHESIS`. If the map itself is the main result, use `BIBLIOMETRICS`. Never resolve conflicting values by guessing. Ask a focused question or mark the item unresolved. ## 2. Previous-letter permission and benchmark extraction Ask whether a previous cover letter exists and how it may be used: - `FORMAT_ONLY`: copy layout properties only; - `FORMAT_AND_TONE`: learn layout and general voice, not distinctive wording; - `MAXIMUM_SUITABLE_WORDING`: retain suitable language only after rechecking every fact and journal detail; - `FACT_CHECK_ONLY`: use solely to identify facts that must be verified elsewhere; - `ANONYMOUS_EXPERT_BENCHMARK`: compare editorial decisions without copying private facts or wording; - `NONE`: do not inspect or reuse it. When benchmark analysis is allowed, extract two independent strengths: 1. `benchmark_empirical_signal`: which concrete manuscript facts, trends, or contrasts the human considered worth foregrounding; 2. `benchmark_editorial_logic`: how the letter converted selected evidence into a reason for editorial action. 3. `benchmark_selection_granularity`: how much manuscript-native specificity was retained (for example, a six-domain taxonomy, three frontier terms, or a directional transition). Do not reduce benchmark analysis to sentence length, tone, or lexical similarity. Learn the human's selection intelligence, selection granularity, and editorial reasoning. The optimized letter should preserve the strongest factual signal at the minimum granularity needed to retain the manuscript's identity while supplying the clearest defensible meaning. For a blind benchmark test, follow `references/blind-benchmark-loop.md`. The benchmark must be inaccessible during baseline drafting and may be revealed only after the baseline is frozen. Never carry over names, titles, journal details, identifiers, metadata, declarations, or unsupported claims. Treat the manuscript and author-confirmed materials as the fact authority even when the benchmark was written by an expert. When a benchmark title, number, date, article label, or declaration conflicts with the current manuscript, record the conflict, learn only the selection logic, and use the verified current fact. ## 3. Author confirmation gate Before journal research or final drafting, establish a compact confirmation record containing: - the sourced fact sheet and unresolved items; - exact `official_article_type` and selected `intellectual_route`; - `empirical_anchor` and `editorial_meaning`; - up to six candidate contributions; - up to three route-specific decision spines or theses; - declaration gaps and likely overclaim risks; - the recorded permission for prior material. Use `EXPLICIT_CONFIRMATION` when the user reviews that record. Use `EVIDENCE_COMPLETE_FAST_PATH` when the user has already supplied unambiguous instructions and every decision-changing fact is verified. Ask only when a conflict, missing declaration, unclear official article type, unclear route, prior-letter permission issue, or other unresolved item could materially change the letter. ## 4. Current journal research After the factual foundation is stable, check current official journal or publisher sources for: - aims, scope, readership, and accepted article types; - the exact article-type label available in the target section; - cover-letter instructions and submission-system constraints; - mandatory declarations; - editor identity only when useful and verified; - recent relevant articles or editorials when they materially sharpen fit. Keep `official_requirement` separate from `reasoned_fit_inference`. Do not rename the submission type to match the intellectual route. Build journal fit in this order: (1) explicit article-type or editorial-priority criterion, (2) concrete readership need, then (3) current journal conversation when useful. Complete a `journal_fit_bridge`: connect one verified scope or readership need to one specific manuscript contribution and explain the decision or understanding this gives the journal's readers. Generic fit language does not pass. If official information cannot be checked, return `NEEDS_JOURNAL_VERIFICATION`, not `SUBMISSION_READY`. ## 5A. Original Research route Build one `research_decision_spine`: > Stakes -> unresolved limitation -> study response -> empirical anchor -> synthesized finding -> editorial meaning -> journal fit Rules: 1. Foreground the specific problem that makes the study worth attention. 2. Define the inference, measurement, or decision that current evidence cannot resolve. 3. Translate methods into capabilities rather than cataloguing tools. 4. Select one concrete finding anchor and one coherent conclusion. 5. State one immediate, defensible consequence. 6. Connect that consequence to a verified journal criterion or readership need. Use contrast only when supported. Preserve design boundaries through accurate verbs. ## 5B. Review and synthesis route Set: ```text synthesis_intervention: what the review does to existing evidence editorial_thesis: the memorable, evidence-supported interpretation synthesis_contrast: the field reading that the synthesis confirms, bounds, or redirects ``` Build: > Field problem -> synthesis intervention -> empirical anchor -> editorial thesis -> editorial meaning -> changed decision or research agenda -> journal fit -> calibrated boundary Use one thesis and at most two supporting cross-study patterns. Methods, registration, search coverage, appraisal tools, and included-study counts establish credibility; they are not the intellectual contribution by themselves. Do not manufacture controversy for a descriptive or scoping review. ## 5C. Bibliometrics route Use this route when the manuscript's principal contribution is field-level mapping from publication metadata, citation relations, term relations, collaboration networks, or temporal patterns. Set: ```text bibliometric_mode: PERFORMANCE_ANALYSIS | SCIENCE_MAPPING | BOTH performance_analysis_signal: the one output, distribution, concentration, or impact pattern that matters science_mapping_signal: the one conceptual, intellectual, or collaboration structure that matters mapping_intervention: what previously invisible field structure or evolution the analysis makes visible mapping_thesis: the one memorable field-level interpretation bibliometric_signature_packet: the minimum manuscript-native combination of structural map, frontier signal, and/or directional transition that preserves the paper's empirical identity authorial_specificity_floor: at least one named or quantified manuscript-native result beyond corpus size bibliometric_decision_spine: field-scale uncertainty -> bounded corpus -> mapping capability -> empirical anchor -> mapping thesis -> coordination or research consequence -> journal fit -> metric boundary metric_boundary: the database, time, metric, and inferential limits that constrain the interpretation ``` The letter should answer four questions: 1. What field-level uncertainty prevents researchers or clinicians from orienting their next work? 2. What structure, transition, concentration, fragmentation, or underconnection does the map reveal? 3. Why does that pattern change research coordination, validation priorities, collaboration, or topic selection? 4. Which metric boundary must be retained so that the map is not mistaken for quality, causality, or prediction? Use one primary map-level conclusion and at most two supporting patterns. For `BOTH`, retain at least one performance-analysis signal and one science-mapping signal, and make both support the same mapping thesis. Do not let a country ranking become the thesis, and do not abstract away the map. Build a `bibliometric_signature_packet` from up to three components: (1) a structural taxonomy or cluster map, (2) a temporal/frontier signal, and (3) a directional transition. Retain two components when they are central. If a taxonomy of three to six clinically meaningful domains is itself a principal result, name the domains in compressed parallel syntax rather than replacing them with 'several themes.' Publication counts, software names, centrality values, and country rankings remain credibility details unless they directly carry the interpretation. Do not equate: - publication or citation volume with research quality; - centrality with causal influence; - co-occurrence with conceptual validity; - a database-indexed corpus with the complete field; - citation bursts with certain future trends; - output concentration with clinical or scientific leadership. Prefer `maps`, `shows concentration`, `identifies a thematic transition`, `reveals fragmentation`, `indicates emerging attention`, or `defines a coordination gap`. Use `predicts`, `proves`, `dominates`, `most influential`, or `future direction` only when the manuscript and metric design directly justify them. A bibliometric letter fails when it is only a catalogue of databases, software, countries, institutions, journals, clusters, or keywords. It also fails when abstract interpretation makes the manuscript empirically interchangeable with any other bibliometric paper. The map must be converted into a field-level decision while retaining at least one named or quantified manuscript-native result. ## 6. Evidence-to-meaning gate Before drafting each substantive paragraph, record: ```text empirical_anchor: source-traceable fact or pattern editorial_meaning: defensible implication for understanding or decision claim_zone: FACT | INTERPRETATION | EDITORIAL_SIGNIFICANCE ``` No free-floating abstractions. No uninterpreted fact lists. If the anchor changes, regenerate the meaning; if the meaning cannot be traced, weaken or remove it. ## 7. Persuasion without overclaiming Use three zones: - scientific facts: no promotional uplift; - study, synthesis, or mapping interpretation: measured strengthening tied to evidence; - editorial significance: the strongest reasonable statement about changed understanding or decisions. Priority claims such as `first`, `only`, `most comprehensive`, `most influential`, or `unprecedented` require independent verification or omission. For bibliometrics, ranking and forecasting language receives additional scrutiny. ## 8. Declarations Verify declarations during intake, but include one in the letter only when: 1. the journal explicitly requires it there; 2. it is standard and decision-relevant for the submission branch; 3. the user requests it; or 4. an unusual circumstance must be disclosed. Do not automatically insert funding, ethics, consent, data/code, preprint, or AI-use statements merely because they were verified. Keep APC or publication-fee willingness, open-peer-review choices, reviewer suggestions, and routine submission-system confirmations out of the letter unless the journal explicitly requires them in the cover letter. Never invent missing declarations. ## 9. Bounded revision loops Maintain at least: ```text official_article_type intellectual_route submission_branch fact_status previous_letter_permission confirmation_mode empirical_anchor editorial_meaning research_decision_spine or editorial_thesis or mapping_thesis synthesis_intervention and synthesis_contrast for Review bibliometric_mode, performance_analysis_signal, science_mapping_signal, mapping_intervention, bibliometric_decision_spine, and metric_boundary for Bibliometrics journal_conversation journal_fit_basis journal_fit_bridge controlled_uplift_level hard_gate_failures quality_gate_failures draft_round length_round stop_reason ``` Run: 1. an intake and route-stability loop; 2. a route-selection loop with no more than three alternatives; 3. a draft-audit-targeted-revision loop with no more than three rounds; 4. an omission and length loop with no more than two rounds. Revise failed dimensions only. With an authorized benchmark, preserve the strongest `benchmark_empirical_signal` and its necessary `benchmark_selection_granularity`, then improve the transferable editorial rule responsible for any material weakness. Do not imitate wording. In blind mode, regenerate from the manuscript and revised skill only; do not place benchmark wording in the generation prompt. ## 10. Final gates and delivery Before delivery, check: - every factual statement is traceable; - current official journal information was verified; - `official_article_type` is exact and does not conflict with the submission system; - the intellectual route matches the primary evidence object; - the empirical anchor and editorial meaning are both recoverable after one reading; - the authorial empirical fingerprint remains visible and the draft is not interchangeable with another paper on the same topic; - route-specific spine or thesis is coherent; - claim strength remains within the design or metric; - journal fit maps a verified criterion or readership need to the contribution; - previous material was used only as permitted; - required declarations are complete; - no placeholders, stale journal details, abstract-like lists, or metric catalogues remain. Deliver: 1. clean final cover-letter text; 2. DOCX when requested or feasible; 3. a concise audit of facts, journal evidence, route, evidence anchor, editorial meaning, and claim strength; 4. items still requiring author confirmation; 5. exactly one final status. Allowed statuses: - `SUBMISSION_READY`; - `NEEDS_AUTHOR_CONFIRMATION`; - `NEEDS_JOURNAL_VERIFICATION`; - `BEST_SAFE_DRAFT_WITH_UNRESOLVED_ITEMS`. Never use `SUBMISSION_READY` when official journal information was unavailable, a fact conflict remains, declarations are incomplete, the official article type is unresolved, or prior-letter use exceeded permission. ## Bundled tools and optional detail - `references/` provides route playbooks, blind-benchmark rules, benchmark rules, and audit rubrics; - `scripts/extract_docx_content.py` extracts DOCX text; - `scripts/validate_payload.py` validates structured output; - `scripts/audit_cover_letter.py` checks placeholders and risky wording; - `scripts/generate_audit_report.py` creates the separate audit; - `scripts/render_cover_letter_docx.py` produces a metadata-scrubbed DOCX; - `assets/` contains templates and example payloads.