Habilidades teardown Historial de auditorías
📦

Historial de auditorías

teardown - 1 auditoría

31 oct 2026, 14:11

No evidence found of keylogging, credential theft, or prompt injection in the reviewed context; most alerts match benign syntax and metadata. Confirmed risks include URL-controlled caption injection and unrestricted recursive deletion through the contact-sheet name. The 394 omitted static matches remain unreviewed and require manual review before automatic publication. Static review was capped at 400/794 representative findings; omitted static matches are unconfirmed, so automatic publishing stays disabled until manual review.

58
Archivos escaneados
9,645
Líneas analizadas
8
Elementos de revisión
0
Falsos positivos ignorados
Elementos de revisión de capacidades (3)

Estas son capacidades locales reales que pueden esperarse para esta habilidad, por lo que requieren revisión, pero no se cuentan como comportamiento malicioso confirmado.

Alto
innerHTML assignment (XSS risk)
storyEl.innerHTML = `<div class="cap ${cur.kind || 'hero'}">
runtime.js merges URL-supplied story configuration into captions. Unescaped tag, text, and kind values reach innerHTML, allowing scripts through injected HTML event handlers.
Alto
Node.js fs operations
fs.rmSync(dir, {recursive: true, force: true});
The unchecked --name value reaches recursive, forced deletion through path.join(base, name). Parent segments can delete directories outside the sheet output folder.
Medio
Node.js fs operations
fs.mkdirSync(dir, {recursive: true});
The unchecked --name also controls this directory and subsequent frame writes. Parent segments permit output creation outside the intended sheet folder.

Factores de riesgo

🌐 Acceso a red (36)
⚙️ Comandos externos (50)
🔑 Variables de entorno (20)
📁 Acceso al sistema de archivos (50)
⚡ Contiene scripts (42)
Auditado por: codex