The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the shodan library via pip and interacts with the well-known api.shodan.io service to retrieve host data, search results, and statistics.
[COMMAND_EXECUTION]: Includes instructions for initializing the Shodan CLI with shodan init and uses sudo pacman for system package installation on specific distributions.
[PROMPT_INJECTION]: The skill processes untrusted data from external sources, creating an indirect prompt injection surface.
Ingestion points: Service banners, HTTP titles, and HTML content retrieved from the Shodan API (SKILL.md).
Boundary markers: The skill includes a mandatory confirmation gate for user authorization, but does not implement specific delimiters to separate untrusted banner data from the agent's internal prompt instructions.
Capability inventory: CLI command execution, network requests, and local file writing (shodan download, shodan parse).
Sanitization: No explicit sanitization, filtering, or escaping of ingested banner content is implemented before processing.

shodan-reconnaissance