About NoAtMark

Why we built it

Copying from ChatGPT, Claude, or the modern web quietly drags in invisible characters — zero-width spaces, joiners, BOMs, variation selectors. These hidden bytes break string matching, corrupt CSV and JSON, trigger spam flags, and confuse AI search engines. Most people never see them until a bug costs them hours.

NoAtMark exists to make text hygiene a one-click habit: paste or upload a file, see what's hidden, and clean it — entirely in your browser.

How the tools work

Every tool is built on small, deterministic engines (scan, clean, plaintext, watermark, inspect) that run locally on your device. No text or files are uploaded. The same engines power the web tools, the JSON API, and the browser/VS Code extensions, so behavior is consistent everywhere.

Accuracy & verification

• Deterministic. Invisible-character detection is based on the Unicode Standard code points — not guesses. Our engines are tested against real copied content.
• Honest scope. We clean invisible characters and formatting artifacts. We do not claim to remove statistical AI watermarks (that requires a model provider's private key) or to bypass detectors. Read why →
• Transparent scores. The Pre-Publish Inspector GEO score explains exactly why it is what it is — every sub-score lists its reasons.

What we follow

• Unicode character semantics for detection (Unicode Standard).
• Content-provenance standards (C2PA) for image transparency.
• Public transparency rules (EU AI Act Article 50) — which is why our pages explain marks rather than help evade them.

Trust & transparency

• Privacy. Nothing you process is collected or stored. See the privacy policy.
• Openness. The core engines are published on GitHub, on PyPI, and on npm.
• No dark patterns. We don't overclaim, we don't hide our limitations, and we don't make guarantees search engines can't back.

Contact

Questions, feedback, or bugs: [email protected].