Este informe no está traducido al idioma solicitado. Se muestra en su lugar el informe canónico en inglés.

Evaluación de seguridad versionada

ID del informe: SA-F93E9BB0

7/7/2026, 5:35:21 AM

ralph evaluación de seguridad v4

Informe de certificación de seguridad de la habilidad

Historial de auditorías
Modelo de auditoría: codex Último informe publicado
Nombre de la habilidad
ralph
Versión
v3.0.0
Mantenedor
supercent-io
Fuente
github.com/supercent-io/skills-template/tree/main/.agent-skills/ralph/
Cobertura
3 Archivos escaneados · 935 Líneas analizadas
Versión de la política
No disponible

Gravedad máxima de hallazgo confirmada

Crítico

2 hallazgos de seguridad confirmados requieren atención.

Contexto de instalación

Consulta la página actual de Skill

Esta página resume únicamente las pruebas del informe. La página de la Skill proporciona el aviso de instalación canónico.

Abrir la página actual de la habilidad

Este informe no bloquea ni autoriza el manifiesto ni el ZIP.

Most external-command findings are false positives caused by Markdown backticks, fenced examples, or literal prompt text. The confirmed risks are persistent writes to hidden agent configuration, Codex developer instruction injection, and broad Gemini hook guidance that can change agent behavior beyond one task.

Posición del informe

Último informe publicado

«Más reciente» se refiere a la secuencia del informe, no a la actualidad del artefacto.

Atestación de auditoría

No se puede certificar

La vinculación inmutable requerida está incompleta.

Verificación humana

No verificado

No se ha registrado ninguna verificación humana para este informe.

Cobertura

3 Archivos escaneados · 935 Líneas analizadas

26 elementos mostrados para revisión

Limitaciones

Este informe no afirma ninguna ejecución en tiempo de ejecución ni en entorno aislado y no prueba la ausencia de efectos secundarios.

Cadena de evidencias

Sigue la evidencia desde la vinculación del código fuente hasta el contrato de instalación. La evidencia disponible permite la verificación; no constituye una garantía de seguridad.

  1. Fuente

    Vinculado al commit y la ruta

  2. Artefacto

    Hashes de contenido y árbol vinculados

  3. Auditoría

    Completo

  4. Contrato de instalación

    Abrir el manifiesto para verificar

    Abrir el manifiesto

Capacidades observadas

Observado significa que este informe registró evidencia de respaldo. No registrado no prueba que una capacidad esté ausente.

Contiene scripts

Puede ejecutar código incluido con la habilidad.

No registrado por esta auditoría

Acceso a red

Puede conectarse a servicios externos.

Observado en 6 ubicaciones de evidencia

Acceso al sistema de archivos

Puede leer o escribir archivos locales.

Observado en 18 ubicaciones de evidencia

Variables de entorno

Puede leer valores del entorno del proceso.

No registrado por esta auditoría

Comandos externos

Puede invocar comandos o programas fuera de la habilidad.

Observado en 155 ubicaciones de evidencia

Elementos de revisión de capacidades (24)
Alto
scripts/setup-codex-hook.sh:5
Hidden file in home directory
# 1) developer_instructions → ~/.codex/config.toml (ooo command contract)
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:6
Hidden file in home directory
# 2) ~/.codex/prompts/ralph.md (load via /prompts:ralph)
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:7
Hidden file in home directory
# 3) ~/.codex/prompts/ouroboros.md (load via /prompts:ouroboros)
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:29
Hidden file in home directory
echo " 1. Adds ooo command contract to ~/.codex/config.toml developer_instructions"
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:30
Hidden file in home directory
echo " 2. Creates ~/.codex/prompts/ralph.md for /prompts:ralph"
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:31
Hidden file in home directory
echo " 3. Creates ~/.codex/prompts/ouroboros.md for /prompts:ouroboros"
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:5
Hidden file access
# 1) developer_instructions → ~/.codex/config.toml (ooo command contract)
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:6
Hidden file access
# 2) ~/.codex/prompts/ralph.md (load via /prompts:ralph)
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:7
Hidden file access
# 3) ~/.codex/prompts/ouroboros.md (load via /prompts:ouroboros)
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:29
Hidden file access
echo " 1. Adds ooo command contract to ~/.codex/config.toml developer_instructions"
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:30
Hidden file access
echo " 2. Creates ~/.codex/prompts/ralph.md for /prompts:ralph"
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:31
Hidden file access
echo " 3. Creates ~/.codex/prompts/ouroboros.md for /prompts:ouroboros"
The line is descriptive text, but it names the hidden Codex home configuration targets that this setup script later creates or writes. Persisting agent prompts and developer instructions in ~/.codex is a real configuration risk.
Alto
scripts/setup-codex-hook.sh:42
Hidden file access
CODEX_DIR="$HOME/.codex"
This defines ~/.codex as the target directory and the script later writes config and prompt files under it. Hidden home-directory agent configuration can persist behavior across sessions.
Alto
scripts/setup-codex-hook.sh:119
Python file write/append
with open(path, "w") as out:
The Python write appends or rewrites Codex developer_instructions in ~/.codex/config.toml. Persistently changing agent instructions is a high-impact configuration risk.
Alto
scripts/setup-codex-hook.sh:125
Python file write/append
with open(path, "a") as out:
The Python write appends or rewrites Codex developer_instructions in ~/.codex/config.toml. Persistently changing agent instructions is a high-impact configuration risk.
Alto
scripts/setup-codex-hook.sh:141
Python file write/append
with open(path, 'a') as f:
The Python write appends or rewrites Codex developer_instructions in ~/.codex/config.toml. Persistently changing agent instructions is a high-impact configuration risk.
Alto
SKILL.md:459
Hidden file in home directory
Required in `~/.gemini/settings.json`:
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.
Alto
SKILL.md:464
Hidden file in home directory
"includeDirectories": ["~/.gemini/extensions/ralph"]
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.
Alto
SKILL.md:469
Hidden file in home directory
AfterAgent hook for loop continuation (add to `~/.gemini/settings.json`):
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.
Alto
SKILL.md:477
Hidden file in home directory
"command": "bash ~/.gemini/hooks/ralph-check.sh",
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.
Alto
SKILL.md:459
Hidden file access
Required in `~/.gemini/settings.json`:
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.
Alto
SKILL.md:464
Hidden file access
"includeDirectories": ["~/.gemini/extensions/ralph"]
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.
Alto
SKILL.md:469
Hidden file access
AfterAgent hook for loop continuation (add to `~/.gemini/settings.json`):
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.
Alto
SKILL.md:477
Hidden file access
"command": "bash ~/.gemini/hooks/ralph-check.sh",
The documentation instructs users to modify hidden Gemini configuration or run hooks from ~/.gemini. This is user-directed, but it expands persistent agent hook and shell execution surface.

Hallazgos de riesgos

Las preocupaciones de seguridad confirmadas están separadas de los elementos que aún deben revisarse.

Preocupaciones de seguridad confirmadas (2)

RISK-001 Crítico
scripts/setup-codex-hook.sh:65-86scripts/setup-codex-hook.sh:97-146
Prompt Injection Attempt Detected
The setup script appends a new Codex developer_instructions contract with phrases like "Ouroboros specification-first workflow is active" and "continue immediately". This persists agent behavior outside the skill session.
The file directly defines the persistent instruction text and writes it into ~/.codex/config.toml. This is a clear agent-instruction persistence mechanism.
RISK-002 Alto
SKILL.md:469-487
Wildcard AfterAgent Hook With Reduced Confirmation
SKILL.md instructs Gemini users to add an AfterAgent hook with matcher * that runs a bash command from ~/.gemini, then suggests sandbox plus yes mode to reduce prompts.
The hook configuration and reduced-confirmation command are shown together in the documented install flow. This creates broad automatic command execution risk if followed.

Medidas correctivas

Esta auditoría registra correcciones sugeridas. Aplicarlas es responsabilidad de la persona encargada del mantenimiento.

  1. FIX-001
    Crítico
    Persistent Codex developer instruction modification
    Do not append to ~/.codex/config.toml automatically. Require explicit consent, show a diff, create a backup, and provide an uninstall path.
  2. FIX-002
    Alto
    Global hidden prompt and hook files
    Prefer project-scoped configuration. Avoid writing reusable prompts or hook commands into hidden home directories without clear user approval.
  3. FIX-003
    Alto
    Wildcard agent hook with reduced confirmation
    Avoid matcher * hooks for shell commands. Require explicit trigger conditions and do not recommend yes mode to suppress confirmations.
  4. FIX-004
    Bajo
    Remote install references are not pinned
    Pin external install instructions to reviewed releases or commits and document how users can verify the source before installing.

Evidencia experta

Identidad inmutable del sujeto, metadatos del escáner, coincidencias descartadas y evidencia a nivel de código fuente.

Sujeto del artefacto

Commit de Marketplace
f93e9bb0daca99badb6a7e574b97737155d57cb3
Hash del contenido
8eb0294915a893556e8755524be7448b36fa3e1f4358653158a9f7557d41c20e
Hash del árbol
3f130ecda94f3c2dbfff596beea431e8787421bf04e49ee45daed6a3f629ef85
Ruta del Skill
skills/supercent-io/ralph
Hash de la carga de auditoría
0a0788d5dd25ce546de6b61275a2e690

Metadatos del análisis

Modelo de auditoría: codex

Estado del análisis: Completo

El alcance se limita a los archivos, líneas, métodos y pruebas registrados. No se afirma ninguna ejecución en tiempo de ejecución ni en entorno aislado.

Verificar y exportar

Informe JSON CITATION.cff SARIF 2.1.0 Manifiesto de instalación firmado Fuente

El manifiesto y el archivo de bloqueo vinculan los artefactos de instalación a hashes criptográficos. Esta afirmación de integridad es independiente de la evaluación de seguridad.

Atestación de auditoría: not_attestable