The Agent Skills Directory

[DATA_EXFILTRATION]: The skill exposes local environment information through hardcoded absolute file paths and an internal network IP address.
Evidence includes the local path: /home/h/Documents/AmericanEmbedded/AmericanEmbedded/VL53L9CX_Pi_Cam.
Evidence includes the internal gallery URL: http://100.64.84.41:8765/.
[COMMAND_EXECUTION]: The skill instructs the agent to execute specific shell commands and scripts to perform exports and rendering.
It runs kicad-cli pcb export step and kicad-cli pcb export glb.
It executes a local shell script found in the project: build/product-renders/export_board_3d.sh.
[DYNAMIC_EXECUTION]: The skill executes a Python script within Blender that is part of the project's directory structure.
Script path: build/product-renders/build_product_scene.py.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes external files and executes scripts from the project root without explicit sanitization.
Ingestion points: Reads .kicad_pcb files and executes scripts (.sh, .py) from the project directory.
Boundary markers: None identified; there are no instructions to ignore embedded commands in the source files.
Capability inventory: Full shell access for kicad-cli and arbitrary Python execution via Blender scripts.
Sanitization: No validation or sanitization is performed on the content of the project files or the scripts before execution.

pcb-product-render