What this scanner does

The Residual Logic Scanner performs five checks against a brief-and-draft pair: Framing, Structure, Assumptions, Narrowing, and Causal. Each check identifies passages in the draft that reflect reasoning the model introduced — reasoning that the author did not specify.

The scanner does not evaluate whether the AI reasoning is correct or appropriate. It surfaces what was introduced so that the author can make an informed decision about whether to retain, revise, or remove it.

Severity levels

Analysis modes

Drafting — Used before editing begins. Identifies what the model introduced into the first draft relative to the brief. Produces a Residual Logic Report.

Traceability — Used after the author has edited the draft. Compares the first draft findings against the final document to show what was resolved, partially addressed, or persisted into the final version. Produces the audit record for retention.

Governance reference

Privacy

• Scan content is not stored by Reality and Reason.
• Text is transmitted to Anthropic's API and processed under Anthropic's data retention policy.
• We store registration details (name, email, organisation, role) and scan count only.
• No request bodies are logged by the server.

Reality and Reason

The scanner is a product of Reality and Reason, publishers of governance instruments for AI-assisted knowledge work. hello@realityandreason.org