The Agent Skills Directory

[SAFE]: The skill consists of instructional content and code examples for the CrewAI multi-agent framework. No malicious patterns, obfuscation, or unauthorized data access were found in the provided instructions or scripts.
[PROMPT_INJECTION]: The skill describes patterns for agents to ingest external data via tools like SerperDevTool and WebsiteSearchTool. This creates a potential surface for indirect prompt injection (Ingestion points: external search/web tools; Boundary markers: not present in examples; Capability inventory: file/web access tools; Sanitization: not present in examples), which is a standard characteristic of the documented framework's intended use for research tasks.
[COMMAND_EXECUTION]: The skill includes a code example for a custom tool that executes SQL queries. This is provided to illustrate the tool-building API within the CrewAI framework and follows standard development practices for extending agent capabilities.

crewai