From fe132e9e96f5293ff113b220b5237a339c625cd5 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Dieter=20Schl=C3=BCter?= Date: Sat, 13 Jun 2026 00:32:31 +0200 Subject: [PATCH] =?UTF-8?q?docs:=20KI-Score-Gewichte=20in=20Konfigurations?= =?UTF-8?q?referenz=20erg=C3=A4nzt?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Abschnitt 12 (Score-Gewichte) listete die ai_*-Punktwerte nicht. Nun vollständig, mit Hinweis auf die Gelb-Deckelung. Co-Authored-By: Claude Opus 4.8 --- BEDIENUNGSANLEITUNG.md | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/BEDIENUNGSANLEITUNG.md b/BEDIENUNGSANLEITUNG.md index 3422bd0..d65d04f 100644 --- a/BEDIENUNGSANLEITUNG.md +++ b/BEDIENUNGSANLEITUNG.md @@ -527,6 +527,18 @@ Alle Einstellungen in `config.yaml`. Fehlende Werte werden aus den Defaults übe | Neuer kaputte Link (4xx/5xx, nicht in Baseline) | 20 | | Geänderte Bild-/Binärdatei (`check-assets`) | 20 | +KI-Inhaltsanalyse (nur wenn aktiviert; Gesamt-Level aus KI ist auf **Gelb gedeckelt**): + +| Ereignis (KI) | Standard-Punkte | +|---|---| +| Pornografie/sexuell expliziter Inhalt (`ai_pornography`) | 50 | +| Diffamierende/strafbare Aussage (`ai_defamation_illegal`) | 50 | +| Politische Propaganda (`ai_propaganda`) | 40 | +| Versteckter Spam (`ai_hidden_spam`) | 40 | +| Problematischer Bildinhalt — Untergrenze (`ai_suspicious_image`) | 40 | +| Themenfremde Werbung (`ai_off_topic_commercial`) | 30 | +| Widersprüchliche Aussage (`ai_contradiction`) | 20 | + Alle Werte sind in `config.yaml` unter `scoring:` anpassbar. ### Alarm-Schwellen (`thresholds:`)