The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided hardware design files and project documentation, which serves as a potential surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: The skill is instructed to load and analyze hardware/*.kicad_sch, hardware/*.kicad_pcb, docs/design-constraints.json, and docs/component-selections.md as context for validation.
Boundary markers: The instructions do not define clear delimiters or provide "ignore embedded instructions" warnings for the data extracted from these files (e.g., schematic notes, net labels, or component descriptions).
Capability inventory: The agent has file Write access to docs/validation-report.md, WebSearch capabilities, and specialized Model Context Protocol (MCP) tools for PCB analysis (mcp__kicad-pcb__*) and manufacturing interaction (mcp__jlc__*).
Sanitization: There is no evidence of sanitization or validation logic applied to the content of the hardware files before it is processed by the agent.

eda-drc