The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and parse untrusted data streams, such as radio captures, hex dumps, and responses from remote telecommand/telemetry services, as part of its core CTF-solving functionality.
Ingestion points: Data enters the agent's context through socket.recv() calls in the Python harness and via file ingestion of IQ/WAV recordings.
Capability inventory: The skill uses network sockets and subprocess calls to specialized SDR and orbital mechanics tools.
Sanitization: The harness uses specific regex patterns (e.g., rb"HTB\{[^}]+\}") to extract flags from untrusted payloads, focusing on protocol-specific parsing rather than generic content execution.
Assessment: This surface is an inherent requirement of the skill's purpose (CTF solving) and is considered safe within the context of an isolated lab environment.
[EXTERNAL_DOWNLOADS]: The skill references several well-known and trusted third-party Python libraries and specialized tools required for aerospace security research.
Dependencies: References include spacepackets, ccsdspy, skyfield, sgp4, pyorbital, gr-satellites, and gps-sdr-sim.
Assessment: These are established tools from reputable sources in the SDR and satellite engineering communities.
[COMMAND_EXECUTION]: The skill provides Python code snippets for network communication and command-line examples for SDR tools (e.g., hackrf_transfer).
Evidence: Templates in ccsds-frame-construction.md and ground-systems-and-rf.md facilitate interaction with remote challenge endpoints.
Assessment: The commands are standard for technical security competitions and do not exhibit malicious intent.

satellite-ctf