parallel-execution
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides templates for spawning subagents that incorporate project context and task descriptions into their system-level instructions. This creates a surface where malicious content within the files or context being analyzed could potentially influence the subagent's actions.
- Ingestion points: The subagent prompt template in
SKILL.mdincludes placeholders for[RELEVANT CONTEXT ABOUT THE CODEBASE/PROJECT]and[CLEAR DESCRIPTION]. - Boundary markers: The provided examples do not use delimiters (like XML tags or triple quotes) or specific "ignore instructions in context" warnings to separate user-provided or project-provided data from the subagent's core instructions.
- Capability inventory: The subagents are granted capabilities to implement modules (e.g., authentication), modify files, and perform security or architecture reviews based on the provided plan.
- Sanitization: The execution protocol does not mention sanitizing, filtering, or validating the context data before it is interpolated into the subagent's prompt.
Audit Metadata