Manuscript Signal Layer

“READ BEFORE YOU’RE READ.”

Clasr shows you what a manuscript is already signaling before anyone reviews it.

Reading as

Study type

Q-profile

Drop manuscript here

PDF or DOCX · browse

Start Signal Reading

SIGNAL REPORT

See what
Clasr sees.

Structured signals mapped into a traceable report.Non-decisional. Yours.

Select Mode:

READING COMPLETED Quantitative Q1 Author Mode

Clasr Signal Report

Manuscript

Polarization metrics in online discourse

Quantitative · Q1 target · Same manuscript, three reading modes

5critical

6major

4minor

Author Mode

Plain-language findings

This version explains what surfaced, why a reviewer may notice it, and what you could consider before submission.

3 · Methodological visibilityCritical

A load-bearing assumption for the network-based metric is declared but never checked against the dataset.

Why it surfaces: A prerequisite assumption that's stated but untested is one of the first things a careful read looks for.
Consider: Report a graph-stability statistic for the sampling window.

Open Author example

Reviewer Mode

Labels and locations

The explanatory layer is removed. Signal vocabulary, evidence, inference type, and surfacing rationale stay visible in one scan.

Section	Severity	Label	Inference	Evidence	Why it surfaces
3	Critical	NETWORK_STABILITY_ASSUMPTION_DECLARED_NOT_TESTED	Obs.	"Assumes a stable interaction graph"; no stability statistic reported.	A prerequisite assumption is stated but not tested.
4	Critical	CROSS-METRIC_DISCREPANCY_DESCRIBED_AS_CONVERGENCE	Obs.	Affective 0.61, issue-based 0.34, network-based 0.58.	The numerical pattern and prose characterization conflict.
10	Critical	DATA_AVAILABILITY	Obs.	No repository, sharing statement, or on-request clause.	One verification route is absent.

Open Reviewer example

WHAT CLASR SURFACES

The patterns a careful reader would catch.

Clasr makes manuscript behavior visible before submission. No score, no verdict.

ClaimsOverreach, argument drift, conclusion scope
EvidenceMethod visibility, figure/text mismatch, numerical behavior
TraceabilityReproducibility gaps, missing data, absent code

THREE OUTPUT MODES

AuthorPlain-language explanation

ReviewerLabels and locations

AdvisorPriority order

Clasr does not score, predict, rewrite, or impersonate reviewers.

View plans

WHY CLASR IS DIFFERENT

Not a chat answer. A structured reading.

A general chat model

May answer differently each time
Follows the shape of the prompt
Blends explanation with opinion
No fixed report structure
Manuscript handling depends on product and account settings

Clasr

Same eleven sections, same order
Severity calibrated to tier and field
Signals tied to manuscript locations
No score, verdict, or reviewer impersonation
Manuscript deleted once the report is generated

Same underlying technology. The difference is that Clasr is built to read the same way every time, and to forget your manuscript when it is done.

We don’t train AI models on your manuscripts. Ever.

Your work stays private. Manuscripts are processed only to provide Clasr’s reading and analysis features.

Clasr Legal