The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified. The skill is designed to improve agent security by providing middleware for rate limiting and content filtering.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input, creating a potential surface for indirect injection attacks.
Ingestion points: The user_text variable is ingested by the agent in the Runner.run example in SKILL.md.
Boundary markers: None explicitly defined in the example prompt structures.
Capability inventory: The skill enables tool execution, specifically demonstrated with the lookup_order function.
Sanitization: The skill implements the DetectPromptInjection rule to identify and block malicious instructions before the agent processes them.
[EXTERNAL_DOWNLOADS]: The skill references official dependencies from the PyPI registry, including arcjet and openai-agents. These are legitimate libraries required for the stated functionality.

integrate-arcjet-guard-openai-agents-py