The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install the official shodan library from well-known and reputable sources, including PyPI (via pip) and official Arch Linux repositories.
[COMMAND_EXECUTION]: The skill documents numerous shell commands to interact with the Shodan CLI, which is the intended and expected behavior for a reconnaissance tool.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external API responses (Shodan), which could potentially contain untrusted instructions. However, the presence of a 'Mandatory confirmation gate' requiring manual user review and authorization for all probing or data extraction actions mitigates the risk of automated obedience to injected instructions.
Ingestion points: Shodan API query results and host details (SKILL.md).
Boundary markers: Includes a strict mandatory confirmation gate requiring the user to confirm targets and scope before any command execution.
Capability inventory: Shell command execution and Python scripting.
Sanitization: Relies on human-in-the-loop verification of all generated commands and target parameters.

shodan-reconnaissance