od-prototype-card-twitter
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary user input via the
$ARGUMENTSvariable to generate design briefs, which in turn provide instructions for file creation and tool execution. This sequence creates a vulnerability to indirect prompt injection. - Ingestion points: The
$ARGUMENTSplaceholder inSKILL.mdaccepts untrusted data from the user. - Boundary markers: The skill instructions do not utilize delimiters or specific directives to isolate or ignore potentially malicious instructions embedded in the user input.
- Capability inventory: The agent is instructed to use file-editing tools for authoring and to call the
register_open_design_artifacttool, providing a clear path for instructions to affect the filesystem and state. - Sanitization: No sanitization, escaping, or validation steps are defined for the user-supplied content before it is processed by the open-design MCP server tools.
Audit Metadata