The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/research.py utilizes the __import__ function to dynamically load the standard datetime module. While the module name is hardcoded as a string literal, dynamic loading is a technique that can be used to evade static analysis detection of specific module usage.\n- [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by ingesting and processing untrusted data from the internet.\n
Ingestion points: Search results (titles, snippets, and URLs) are retrieved from the web via the BrowserAct API and processed in scripts/research.py.\n
Boundary markers: The skill uses Markdown headers to structure the report, but does not provide explicit boundary markers or instructions to the agent to ignore potential commands embedded within the retrieved snippets.\n
Capability inventory: The skill possesses the ability to perform network requests (via requests) and write files to the local system (via argparse and open).\n
Sanitization: No validation or sanitization is performed on the web content before it is interpolated into the final Markdown report.\n- [DATA_EXFILTRATION]: The skill transmits search queries and authentication tokens to the mcp.browseract.com domain. This communication is required for the skill's primary functionality and targets the vendor's own infrastructure.

web-research-assistant