The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes banner data, HTTP headers, and HTML content from Shodan, which are collected from arbitrary third-party systems. This external data could potentially contain malicious instructions targeted at the agent.\n
Ingestion points: Data retrieved through shodan search, shodan host, and shodan download commands described in SKILL.md and references/detailed-guide.md.\n
Boundary markers: The skill implements a 'Mandatory confirmation gate' requiring the user to verify targets and authorization before any active probing or data extraction.\n
Capability inventory: Shell command execution via the shodan CLI, Python script execution, and local file system writes for data export.\n
Sanitization: The skill relies on manual user verification steps rather than automated content filtering to mitigate injection risks.\n- [COMMAND_EXECUTION]: The skill makes extensive use of the Shodan command-line interface to perform reconnaissance tasks, such as searching for vulnerable services and downloading statistics.\n- [PRIVILEGE_ESCALATION]: Documentation in references/detailed-guide.md mentions the use of sudo pacman for installing the Shodan CLI on specific operating systems, which requires elevated privileges for package management.\n- [EXTERNAL_DOWNLOADS]: The skill fetches infrastructure data and tools from Shodan's official domains. Shodan is a well-known security service, and these operations are essential to the skill's defined purpose.

shodan-reconnaissance