The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes KiCad board, project, and schematic files which could contain malicious instructions embedded in net names or metadata.\n- Ingestion points: The skill reads .kicad_pcb, .kicad_pro, and .kicad_sch (via netlist export) files.\n- Boundary markers: No explicit delimiters or safety warnings are used to wrap data ingested from these files before processing by the agent.\n- Capability inventory: The skill can modify board files via the pcbnew API and execute system commands via the KiCad CLI.\n- Sanitization: Data is parsed using a combination of regex and the official KiCad Python API.\n- [COMMAND_EXECUTION]: The skill executes the kicad-cli binary through subprocess.run() to perform board synchronization and design rule checks.\n- The binary path is sourced from a configurable environment variable KICAD_CLI or common installation paths.\n- Commands are executed using a list of arguments, which prevents common shell injection attacks.

kicad-board-edit