Source quality

1. Click Readiness.

2. Click Source quality.

3. The check looks up retraction and preprint status for cited records (Crossref and related metadata). It is deterministic and does not call the drafting model.

Source quality is not applied automatically on export. Run it before you submit.

AI-use footprint

1. Click Readiness.

2. Click AI-use footprint.

3. The inspector reports how much of the draft carries Rivul's own AI-origin marks, for the whole document and per section. It counts characters Rivul wrote or revised for you, so it is a record of this editor's actions, not a detector. Text pasted in from another tool reads as yours, and a sentence you rewrote by hand over an AI draft can still read as AI. Treat it as an authoring aid, never as evidence you can offer a venue.

Consistency

1. Click Readiness.

2. Click Consistency check.

3. The report covers four mechanical inconsistencies: US and UK spellings mixed in one draft, Fig. against Figure reference styles, an acronym used before it is defined, and spelled-out numbers mixed with digits. It checks style, not values, so it cannot tell you that two sections report different numbers for the same quantity, and it does not compare mathematical notation. Apply a suggested fix only after you read the surrounding sentence.

Citation-worthiness

1. Click Readiness.

2. Click Citation-worthiness.

3. Claim-style sentences without a citation are listed. Insert a Cite marker or rewrite the sentence.

Other lenses

Draft review is the only Review lens that calls a model. It sends the whole draft and its reference list, then asks for a reviewer's critique: unsupported or overstated claims, missing or suspicious citations, contradictions, and clarity, each with the passage quoted and a fix suggested. Because a model writes it, the report varies between runs and is an opinion, not a finding. Counting how many of a sentence's terms appear in its cited passage is a different lens — open Claims then Check claim, described under Claim checking.

Submission readiness searches the draft for the markers reviewers look for (a baseline comparison, formal notation, an ablation, a limitations statement, statistical anchors) and quotes the text that matched. It is a keyword scan, not a review: it does not call a model, score the paper, or recommend a decision. The report arrives as plain text you can insert.

Proofread flags grammar and clarity.

Tone of voice… restyles the selection (or the paragraph at the cursor).

Similarity check (your library) flags near-verbatim overlap between the draft and extracted library text.

Source quality

1. Click Readiness.

2. Click Source quality.

3. The check looks up retraction and preprint status for cited records (Crossref and related metadata). It is deterministic and does not call the drafting model.

Source quality is not applied automatically on export. Run it before you submit.

AI-use footprint

1. Click Readiness.

2. Click AI-use footprint.

3. The inspector reports how much of the draft carries Rivul's own AI-origin marks, for the whole document and per section. It counts characters Rivul wrote or revised for you, so it is a record of this editor's actions, not a detector. Text pasted in from another tool reads as yours, and a sentence you rewrote by hand over an AI draft can still read as AI. Treat it as an authoring aid, never as evidence you can offer a venue.

Consistency

1. Click Readiness.

2. Click Consistency check.

3. The report covers four mechanical inconsistencies: US and UK spellings mixed in one draft, Fig. against Figure reference styles, an acronym used before it is defined, and spelled-out numbers mixed with digits. It checks style, not values, so it cannot tell you that two sections report different numbers for the same quantity, and it does not compare mathematical notation. Apply a suggested fix only after you read the surrounding sentence.

Citation-worthiness

1. Click Readiness.

2. Click Citation-worthiness.

3. Claim-style sentences without a citation are listed. Insert a Cite marker or rewrite the sentence.

Other lenses

Draft review is the only Review lens that calls a model. It sends the whole draft and its reference list, then asks for a reviewer's critique: unsupported or overstated claims, missing or suspicious citations, contradictions, and clarity, each with the passage quoted and a fix suggested. Because a model writes it, the report varies between runs and is an opinion, not a finding. Counting how many of a sentence's terms appear in its cited passage is a different lens — open Claims then Check claim, described under Claim checking.

Submission readiness searches the draft for the markers reviewers look for (a baseline comparison, formal notation, an ablation, a limitations statement, statistical anchors) and quotes the text that matched. It is a keyword scan, not a review: it does not call a model, score the paper, or recommend a decision. The report arrives as plain text you can insert.

Proofread flags grammar and clarity.

Tone of voice… restyles the selection (or the paragraph at the cursor).

Similarity check (your library) flags near-verbatim overlap between the draft and extracted library text.