tdd-workflows-tdd-red
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features an attack surface for indirect prompt injection by interpolating user-supplied data directly into its core instructions without sufficient isolation.
- Ingestion points: The
Prompt Templatesection inSKILL.mduses the$ARGUMENTSplaceholder to incorporate external user input into the agent's instructions. - Boundary markers: Absent. The skill does not utilize delimiters (such as triple quotes, XML tags, or code blocks) to separate the system instructions from the user-provided
$ARGUMENTS, nor does it include explicit instructions for the model to ignore any potentially malicious commands embedded within that data. - Capability inventory: The skill has the capability to write files (generating test files) and encourages the agent to execute them (as seen in the
Validationsection step 1: "Run tests"). This combined capability increases the impact of a successful injection. - Sanitization: Absent. There are no defined filters, validation rules, or escaping mechanisms for the data passed through
$ARGUMENTS.
Audit Metadata