Inspect what is there, decide what needs work, and rewrite with a model you choose. Keep the evidence and the original close.
Reveal named zero-width characters, unusual spaces, direction controls, and selected mixed-script lookalikes at exact positions. Review context, then apply conservative cleanup with one-step undo.
Understand invisible charactersReview repetitive phrasing, familiar framing, and uniform sentence rhythm with contributing passages. The method explains its cues and abstains when the text is unsuitable. These are editing prompts, never proof of AI authorship.
Read about detector limitsSet tone, editing strength, protected terms, and what to avoid. An optional browser-local Voice Reference turns a writing sample into traits you approve; the raw sample and excerpts are not sent as that reference.
Build a useful voice briefReview a token redline, exact source and result artifacts, and checks for numbers, links, quotes, and protected terms. Edit the clean copy and consult saved history. You still review facts and meaning yourself.
Explore the workspaceUse a single pass, a rewrite followed by review, or a fixed three-stage map, draft, and review. The same selected route handles each stage, with the complete maximum shown before you authorize it.
Quick cleanup and straightforward edits.
Strong voice and structure at moderate cost.
Nuanced editorial judgment and fidelity.
Billing and model dispatch are being migrated. Some Studio actions may be unavailable while that work is completed. Model routes and availability can change; the public inspector works independently.
See available and held credit, review wallet activity, and export billing activity as CSV. Every model workflow requires approval of its maximum cost.
How pricing works ↗Studio includes revocable bearer tokens for compatible MCP clients. Inspect, estimate, reserve, advance an explicit stage, and read status through the same bounded workflow.
Explore MCP setup ↗A pattern is a reason to look closer. Your judgment makes the final edit.