ai-social-media-content
Pass
Audited by Gen Agent Trust Hub on May 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
beltCLI tool to interface with various AI models (FLUX, Veo, etc.). Access to this tool is explicitly permitted in the skill configuration. - [EXTERNAL_DOWNLOADS]: The skill documentation includes links to installation scripts and complementary skill modules hosted on the official
inference-shGitHub repository. These resources are integral to the intended functionality of the platform. - [PROMPT_INJECTION]: The skill employs string interpolation to insert variables into AI prompts (e.g., the
$CONCEPTvariable in the repurposing pipeline). This creates an attack surface for indirect prompt injection if the variable content is derived from untrusted sources, although the examples provided use user-defined strings. - Ingestion points: Shell variables within prompt templates in
SKILL.md. - Boundary markers: None present.
- Capability inventory: Capability to execute AI model inference and other shell commands via the
belttool. - Sanitization: No explicit sanitization or filtering of prompt inputs is performed within the provided examples.
Audit Metadata