The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingest data from external web sources, which serves as a surface for indirect prompt injection. This is an inherent risk of its primary research purpose.
Ingestion points: Data is gathered via web_search, fetch_url, and fetch4ai (SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the synthesis steps.
Capability inventory: The skill utilizes write_file, read_file, and task tools to manage findings and subagents (SKILL.md).
Sanitization: No content sanitization or validation of external web data is described.
[COMMAND_EXECUTION]: The skill utilizes command-line operations for environment setup and data extraction.
Creates research directories using mkdir (SKILL.md).
Executes a local utility script fetch4ai.py located in the agent's skills directory (SKILL.md).
[SAFE]: No malicious behavior, such as hardcoded credentials, data exfiltration to suspicious domains, or persistence mechanisms, was detected. The instructions are focused on the stated research task and follow professional practices for data organization.

web-research