The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill contains a verification step that uses curl to check connectivity with crt.sh, a well-known and trusted Certificate Transparency log service. This operation is used solely for environment verification and discards all downloaded data to /dev/null without execution.
[COMMAND_EXECUTION]: Detailed methodology instructions include the use of common security and networking CLI tools such as curl, subfinder, nmap, and masscan. These are intended for the operator or agent to perform reconnaissance against targets within an authorized scope.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process and analyze external, untrusted data (e.g., web pages, JavaScript source code, and API responses). This creates a surface for indirect prompt injection, where an attacker could embed malicious instructions in the target's public assets to influence the agent's logic during an assessment.
Ingestion points: Analyzes external web content, JavaScript bundles (§13), and API schemas (§12).
Boundary markers: Includes high-level instructions to maintain OpSec and avoid pasting sensitive data into cloud LLMs (§5).
Capability inventory: Uses network tools (curl) and file-logging for evidence preservation (§4).
Sanitization: Recommends the use of read-only validators and local LLM models for sensitive data analysis to mitigate exposure risks.

osint-methodology