The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several shell scripts (domain-footprint.sh, verify-domain.sh, and osint-capabilities.sh) to perform legitimate OSINT tasks. These include DNS queries via dig, ownership lookups via whois, and system binary checks. These operations are limited to passive reconnaissance and are consistent with the skill's stated purpose.
[EXTERNAL_DOWNLOADS]: The skill performs passive data retrieval from well-known public intelligence services, including crt.sh (certificate transparency) and the Wayback Machine (archive.org). These connections are used to fetch public records and do not involve the execution of remote code or the download of untrusted scripts.
[PROMPT_INJECTION]: The skill processes external data from websites and APIs, which presents a surface for indirect prompt injection (Category 8). However, the skill provides extensive defensive instructions to mitigate this, such as the 'Four Gates' protocol, which requires the agent to verify claims against multiple independent primary sources and apply strict confidence grading before reporting. Given that data ingestion is the primary purpose of an OSINT tool, this risk is considered managed and acceptable.
[SAFE]: The osint-capabilities.sh script performs an audit of the local environment to detect the presence of specific binaries (e.g., git, python3) and environment variables (e.g., GITHUB_TOKEN, SHODAN_API_KEY). This diagnostic behavior allows the skill to adapt its functionality based on available tools. The script explicitly avoids printing secret values, checking only for presence, which is a safe practice for complex agentic tools.

osint