Technical checks and copyediting for English-language journals
QKCLAW checks incoming manuscripts before peer review, including every reference against Crossref, and copyedits accepted papers in the authors' own Word file. Every change is a tracked change with a comment explaining why. Your editors accept or reject each one.
Drought stress havehas become more and moreincreasingly frequent in the region,; however, field evidence on soil moisture thresholds remains limited [3].
Soil moisture was measured at depths of 10 cm and 30 cm by time-domain reflectometry (TDR) at all four sites.
Grain yield decreased by 18.4% under severe stress (Table 2), and the sensibility of the threshold model was highest at the driest site.
Measured in October 2026 trial runs: technical checks on four English manuscripts (5,900–14,700 words, 22–73 references), copyediting on two of them. Times vary with length, the number of references and network conditions.
Three checks between submission and production
Screen a submission before it goes to reviewers, verify its reference list, and copyedit the accepted version. Each service can be used on its own.
Technical check
Software counts and checks structure, declarations, author details and numbering; the AI assistant writes the assessment, the main issues and comments the editor can send to the authors.
- Input
- Word or PDF manuscript
- Delivered
- Technical check report (Word), English or Chinese
Reference verification
Every reference is looked up in Crossref: does the DOI exist, does it point to the cited work, has the work been retracted. The manuscript's own title is searched against published papers.
- Input
- The manuscript's reference list
- Delivered
- Per-reference findings in the report
Copyediting
Grammar, word choice and consistency are corrected sentence by sentence in the authors' file. Anything that needs the authors' judgement becomes a query, not an edit.
- Input
- Word manuscript
- Delivered
- Tracked-changes Word file with English comments
Software counts the facts, the AI assistant writes the judgement
Word counts, missing declarations and numbering gaps are found by deterministic software, so the numbers in the report can be checked. The AI assistant reads the whole manuscript for the points that need understanding, and every issue points to a paragraph.
Structure and length
- Word count of the main text and abstract
- Abstract, keywords, references, IMRaD sections
- Article type and whether the structure fits it
- Heading numbering gaps and duplicates
Declarations and authors
- Funding, author contributions, conflicts of interest, data availability
- Whether ethics approval and consent apply
- At least one institutional e-mail address; ORCID iDs
- Co-first and corresponding author notes
Figures and citations
- Figures and tables cited in order of first mention
- Citation numbers beyond the reference list
- References never cited in the text
- Each image reviewed for watermarks, consent and quality (Word files)
Integrity signals
- Hidden instructions aimed at AI reviewers
- Similar published papers found by title
- Potentially sensitive or offensive wording
- Language level: ready for review or not
1. Recommendation
Return to the authors for revision before peer review2. Facts counted by the system
- Length
- 6,420 words; main text 4,980
- Structure
- Abstract 243 words; keywords 5
- Declarations
- Not found: data availability, ethics approval
- Authors
- 3 e-mail addresses, none institutional
- Numbering
- Figure 4 is mentioned before Figure 3
3. Main issues
- Paragraph 14: patient records are analysed, but no ethics approval or consent statement is given.
- Paragraph 31: an association from a cross-sectional survey is described as “causes”.
- Reference 27: the DOI resolves neither in Crossref nor at doi.org.
Watermark of an image website, lower right; ask for the source.
Long side under 1,200 px; axis text hard to read.
Bar chart; matches its caption.
Every reference looked up, and “not found” kept apart from “wrong”
The check runs against the public Crossref registry, with doi.org as a fallback for DOIs registered elsewhere. Only what the registries actually show goes into the report; a lookup that failed is reported as not checked, never as fine.
What is checked
- The DOI exists in Crossref or at doi.org
- The DOI points to the cited title
- Retraction and withdrawal notices
- References without a DOI, by bibliographic search
What is reported
- DOIs that resolve nowhere
- DOIs registered to a different work
- Journal-like entries with no record anywhere
- DOIs the authors can add
The manuscript itself
- Its title searched against published papers
- Matches by an author with the same surname listed first
- A same-title paper in your own journal flagged as its published version
Also counted
- Share of references from the last five years
- References sharing a surname with the authors
- Numbered by the list's own numbering, gaps included
| No. | Reference (shortened) | Result |
|---|---|---|
| 4 | Chen L, Wang Y. Field-scale soil moisture monitoring… Agric Water Manag. 2021 | Matched |
| 12 | Okafor N, et al. Drought indices for rain-fed wheat… J Hydrol. 2019 | Retracted 2022 |
| 19 | Smith J. Root water uptake under deficit… 2020;45:113–9. doi:10.5555/rwu.2020.113 | DOI not found |
| 23 | Liu H, Zhang Q. Canopy temperature as a stress index… doi:10.5555/cti.2018.044 | DOI of another work |
| 31 | Garcia M, Rossi P. Threshold models for irrigation scheduling… Field Crops Res. 2022 | Add DOI |
| 36 | Allen RG, et al. Crop evapotranspiration. Rome: FAO; 1998 | Book, not in Crossref |
Edited in the authors' file, every change explained
Changes go into the original Word file as tracked changes, so the layout stays as it was. Comments are in English and can go back to the authors as they are. Where the assistant would have to guess what the authors meant, it asks instead.
Language, sentence by sentence
- Grammar, agreement, tense and articles
- Word choice, informal and non-native phrasing
- Comma splices, fragments, dangling modifiers
- Spelling, missing and repeated words
Consistency
- US or UK spelling, whichever the authors use most
- Hyphenation and the serial comma
- Abbreviations defined at first use
- Number–unit spacing, en dashes in ranges
Queries, not guesses
- Numbers that disagree between sections
- A wrong word whose intended term is unclear
- Names of scales, instruments and guidelines
- Citations that do not fit the statement
Safeguards
- An edit that adds information becomes a comment
- Large rewrites become suggestions, not edits
- Equations and cross-reference fields are left alone
- Suggestions that cannot be located are not dropped
Upload on the web, or simply send an e-mail
No new system to learn. Editors upload a manuscript on the trial site or e-mail it from their registered address, and get the result back the same way.
Upload a Word file (or a PDF for the technical check), or attach it to an e-mail to check@ or edit@.
English manuscripts are counted in words and checked against English-journal conventions; Chinese manuscripts go to the Chinese service.
A minute or two later the report or the tracked-changes file is ready; you are notified by e-mail. Several manuscripts can be queued.
Forward the comments to the authors, or accept and reject changes under Review in Word.
PDFJX-2026-0412-manuscript.pdf
Wtechnical-check-report.docx
One configuration per journal
Required declarations, length limits, report language and house style are written into each journal's own configuration. The same manuscript gets a different report under a different journal.
| Clinical journalexample configuration | Engineering journalexample configuration | |
|---|---|---|
| Report language | English, for an international editorial board | Chinese for the editorial office; comments to authors in English |
| Required declarations | Funding, author contributions, conflicts of interest, data availability, ethics approval and consent | Funding, conflicts of interest, data availability |
| Length | Main text at least 3,000 words for an Article; abstract 150–250 words | Abstract up to 300 words; 3–8 keywords |
| Reference style | Numbered, cited in order of first appearance | Author–year |
| Comments signed as | Tracked changes, comments and the report carry your journal's editorial assistant name | |
How we work
The assistant finds problems and explains them. Decisions stay with the editors.
Where the manuscripts go
Two ways to run QKCLAW, with the same checks.
Frequently asked questions
Will it make mistakes?
Sometimes, which is why nothing is changed outside tracked changes and every change has a reason the editor can reject. Edits that add information or rewrite too much are turned into comments; uncertain words and names become queries.
Which files are supported?
Word (.docx, .doc) for both services, with changes written into the original file. PDF is accepted for the technical check; image-by-image review needs the Word file.
Is this a plagiarism check?
No. The technical check searches published papers by the manuscript's title and verifies the references, but it does not measure text similarity. Keep using your similarity service.
Can it follow our journal's requirements?
Yes. Required declarations, word and abstract limits, reference style, report language and the name used for comments are set per journal.
How long does it take?
In our trial runs, a technical check took 27–62 seconds including reference verification, and copyediting a 5,900–7,400-word paper took 2–2.5 minutes. Longer manuscripts take longer.
Does it handle Chinese manuscripts too?
Yes. The language is detected automatically; Chinese manuscripts get the Chinese precheck and copyediting service described on the Chinese site.
Request a trial
QKCLAW is open to journal editorial offices for a trial, configured for your journal. We first run some of your past manuscripts and show you the results next to what your editors did.
Questions about a configured or installed version: [email protected]
- Send your requirements and a few past manuscripts
- We configure the checks and compare the results
- Start once your editors are satisfied