The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands and scripts to perform PCB exports and rendering tasks. * Evidence: Uses kicad-cli pcb export step and kicad-cli pcb export glb to generate 3D assets. * Evidence: Invokes project-specific shell scripts (build/product-renders/export_board_3d.sh) and Python scripts (build/product-renders/build_product_scene.py) located within the user's workspace.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted project data while possessing the capability to execute commands. * Ingestion points: Reads .kicad_pcb files and DRC reports (build/render-review/drc-latest-render.rpt) to guide the rendering process. * Boundary markers: No specific delimiters or instructions to ignore embedded content are used when processing board metadata or reports. * Capability inventory: The agent can run arbitrary shell scripts (.sh) and Python scripts (.py) found in the build/product-renders/ directory, as well as kicad-cli tools. * Sanitization: No sanitization or validation of the KiCad file content or DRC report text is performed before it influences the agent's workflow logic.
[DATA_EXFILTRATION]: The skill references a local network endpoint for viewing results. * Evidence: Mentions a gallery served at http://100.64.84.41:8765/. This address resides within the IANA-reserved range for Shared Address Space (often used by private mesh networks like Tailscale), indicating a private local deployment rather than public exfiltration.

pcb-product-render