The Agent Skills Directory

[PROMPT_INJECTION]: The skill defines a parallel sub-agent pattern in DESIGN-IT-TWICE.md that ingests user-provided codebase details and technical briefs to prompt multiple sub-agents via the Agent tool. This creates a surface for indirect prompt injection where malicious instructions embedded in the analyzed code could potentially influence the sub-agents' analysis or behavior. \n- Ingestion points: User-provided 'deepening candidate' and 'technical brief' (containing file paths, coupling details, and dependency info) are interpolated into sub-agent prompts. \n- Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when passing external content to sub-agents. \n- Capability inventory: The skill uses the Agent tool to spawn and prompt multiple sub-agents based on the provided technical context. \n- Sanitization: No explicit sanitization or filtering of the user-provided technical content is mentioned before it is processed by the AI pipeline.

codebase-design