skills/jeremylongshore/tons-of-skills-marketplace/ideogram-reference-architecture/Gen Agent Trust Hub
ideogram-reference-architecture
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a prompt templating system in
SKILL.mdthat interpolates external data into model instructions without explicit sanitization or boundary markers. - Ingestion points: The
buildPromptfunction inSKILL.mdaccepts avarsobject containing untrusted strings for interpolation. - Boundary markers: The templates use simple
{subject}placeholders without delimiters or instructions to ignore embedded commands. - Capability inventory: The skill includes functions that perform network operations (
fetchto Ideogram API) and file system operations (writeFileSync,mkdirSync). - Sanitization: The code uses a basic string replacement (
prompt.replace) with no validation, filtering, or escaping of the input variables.
Audit Metadata