The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs shell execution of a provided Python script scripts/md_to_pdf.py. This script is used to consolidate project documentation into a formatted PDF. Analysis of the script source code shows it uses standard libraries (weasyprint, markdown) for document layout and does not contain malicious logic.
[EXTERNAL_DOWNLOADS]: The skill fetches technical documentation (datasheets in PDF format) and CAD footprints from industry-standard websites such as AllDatasheet, LCSC, and official component manufacturer domains. These downloads are essential for the hardware selection and design process.
[PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests untrusted data from external electronic component datasheets and processes them to generate hardware architecture reports.
Ingestion points: Data enters the agent context from external URLs during the datasheet and footprint download phases, and from user-provided project requirements in docs/hardware/01-requirements.md.
Boundary markers: The instructions do not define explicit boundary markers or "ignore" directives for the downloaded content.
Capability inventory: The skill has the capability to execute a local Python script and perform network requests for file downloads.
Sanitization: There is no explicit sanitization or validation logic described for the external PDF or CAD data before processing.

hardware-solution