The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to gather and process data from various external sources including social media bios, WHOIS records, and DNS responses. This introduces an indirect prompt injection surface where malicious content embedded in these public records could influence the agent's behavior during an investigation.
Ingestion points: Data fetched from platform URLs (e.g., GitHub, Twitter, Reddit), DNS records (A, MX, TXT), WHOIS responses, and certificate transparency logs via crt.sh.
Boundary markers: No explicit boundary markers or instructions to ignore embedded prompts are provided in the code snippets.
Capability inventory: The skill utilizes Python network requests (aiohttp), bash commands (dig, whois, subfinder, curl), and interacts with the memory MCP server to store findings.
Sanitization: The provided Python patterns do not include sanitization or validation of the retrieved content before it is processed or stored.
[COMMAND_EXECUTION]: The skill provides templates for executing system commands such as dig, whois, and subfinder to perform network reconnaissance. While standard for OSINT, these commands interact with the local operating system.
[EXTERNAL_DOWNLOADS]: The skill uses curl to fetch JSON data from crt.sh, which is a well-known certificate transparency log service. This is a legitimate source for reconnaissance data.

osint-recon