create-red-green-examples
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions guide the agent to generate code examples and executable verification commands based on user-provided goals and context. This ingestion of untrusted data creates a surface for potential injection attacks.
- Ingestion points: User-provided implementation goals and code snippets used to generate RED/GREEN pairs (SKILL.md).
- Boundary markers: Absent; the skill does not specify delimiters or warnings to ignore instructions embedded in user-supplied data.
- Capability inventory: The skill requires the generation of specific shell commands and assertions (
Check: <exact command>) to verify implementation behavior (SKILL.md, references/pair-construction.md). - Sanitization: Absent; the skill instructions do not include requirements for validating or filtering user-supplied content.
- [DYNAMIC_EXECUTION]: The skill involves the generation of executable code snippets and shell commands from user-influenced templates. While the instructions recommend using isolated workspaces, the generation of executable content based on arbitrary user requests is a documented security surface.
Audit Metadata