The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes KiCad project files (schematics and PCB layouts) which are external, potentially untrusted data sources. This constitutes an attack surface where malicious instructions could be embedded in metadata or design notes. However, the risk is negligible due to the structured nature of CAD files and the specific engineering context of the skill.
Ingestion points: KiCad project files such as .kicad_sch and .kicad_pcb.
Boundary markers: None explicitly defined in the prompt to separate design data from agent instructions.
Capability inventory: The kicad MCP tool provides access to the project filesystem and KiCad's internal design data.
Sanitization: The skill does not define specific sanitization, relying on the underlying kicad-mcp-server for safe data handling.
[COMMAND_EXECUTION]: The skill configuration includes an MCP tool that executes a local Node.js script to interface with KiCad software. This is a functional requirement for the skill and targets a specific local installation path.
Evidence: The mcpConfig field specifies the execution of node targeting ~/.local/share/oh-my-embedded/kicad-mcp-server/dist/index.js.

pcb-designer