START WITH YOUR QUESTION

Find your next manuscript task

Choose what you need to improve. Each task spells out the inputs, the deliverable and a request you can use after installing the skills.

01 / USE CASE

Improve a whole draft

Make the scientific question, argument and evidence work together.

What you provide
Your current draft, relevant results and figures, and the target venue if decided.
What you receive
A scoped revision, the material changes and unresolved evidence gaps.
COPYABLE REQUEST
Use paper-workflow to improve my current manuscript using the supplied results and figures. Start with the scientific argument. Preserve established results, measured values and protected passages. Save a revised copy; explain material changes and unresolved evidence gaps.

Established results and protected passages stay fixed. Missing evidence is identified, not invented.

02 / USE CASE

Rewrite a section

Give the Abstract, Results or Discussion a clear scientific job.

What you provide
The section, its surrounding context, settled Results and your editing scope.
What you receive
A focused revision aligned with the section’s purpose and upstream evidence.
COPYABLE REQUEST
Use paper-workflow to revise only the Abstract of my methods manuscript. Preserve measured values and established Results claims. Flag claims that depend on unsettled experiments. Follow the target journal’s format when provided. Do not modify other sections.

A request to revise the Abstract does not authorize rewriting other sections.

03 / USE CASE

Plan figures & Results

Give every figure one main claim and every panel a reason to be there.

What you provide
The scientific question, result tables, existing figures and available source data.
What you receive
A figure plan with main claims, panel roles and aligned legends.
COPYABLE REQUEST
Use figure-planner to plan the main figures from the supplied results. State the main claim of each figure, what each panel tests and what belongs in the supplement. Align the proposed legends with Results. Give a plan only; do not claim to have rendered or checked figures.

Planning does not render a figure. Production needs the figure skills and a working plotting environment.

04 / USE CASE

Check references & claims

A paper can exist without supporting the sentence that cites it.

What you provide
The claims, their citations, your bibliography and accessible original sources.
What you receive
Separate findings on bibliography consistency and claim-to-source support.
COPYABLE REQUEST
Check the supplied manuscript claims and bibliography with citation-verifier and claim-source-verification. Separate metadata consistency, source existence and whether the original source supports each claim. Report locations and inaccessible sources. Diagnose only; do not edit manuscript files.

Source existence and scientific support are distinct. Inaccessible sources remain unchecked.

05 / USE CASE

Check before submission

Find what would block submission, not another reason to rewrite everything.

What you provide
The final draft, supplementary information, references and target journal.
What you receive
Prioritized issues with locations, remedies and explicitly unchecked categories.
COPYABLE REQUEST
Use submission-audit to inspect the final manuscript, supplementary information and bibliography for the specified journal. Prioritize submission blockers, statistical reporting and evidence consistency. Give locations and fixes. Distinguish checked from unchecked requirements. Do not rewrite the manuscript.

Current journal requirements require accessible sources. A local check is not a complete submission certification.

06 / USE CASE

Respond to reviewers

Connect every reviewer request to an answer and an actual manuscript change.

What you provide
Original numbered comments, the manuscript and completed new evidence.
What you receive
A response organized by the reviewer’s numbering and matched manuscript edits.
COPYABLE REQUEST
Use paper-workflow to prepare a response to the supplied reviewer comments. Preserve every original number and sub-question. Use only completed evidence. Revise the manuscript where authorized, link each response to the actual change and mark requests that remain unresolved.

Planned experiments must not be described as completed results.

Only need a diagnosis?

Say so explicitly: “Identify the main argument problems, rank them by importance and give locations. Do not edit files.” Reviewing and rewriting are different permissions.

Review, survey or Perspective

Build a governing argument and display-item plan using the dedicated Review architecture, rather than forcing an experimental Results structure onto it.

Open the specialist ↗