The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external RF signals, which constitutes a vulnerability surface where maliciously crafted signals could attempt to influence the agent's behavior through decoded outputs.\n
Ingestion points: Signal data is collected via rtl_sdr and hackrf_transfer, and decoded using urh and rtl_433 as described in SKILL.md and references/commands.md.\n
Boundary markers: The skill lacks instructions for implementing boundary markers or 'ignore' directives when the agent interprets the output from tools like rtl_433 (e.g., when using the '-F json' flag).\n
Capability inventory: The skill possesses capabilities for shell command execution, file system writes (.iq and .csv files), and RF transmission via HackRF.\n
Sanitization: There are no documented procedures for sanitizing or validating decoded bitstreams or tool outputs before they are processed by the agent.

re-sdr