The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill guides users to install the arcjet package with the claude-managed-agents extra from PyPI. This is the official library provided by the skill author for security integration.
[COMMAND_EXECUTION]: Includes standard instructions for setting up the development environment using pip install.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted user input (user_text). It addresses this risk by providing a framework to screen messages for prompt injection using DetectPromptInjection before they are sent to the agent session.
Ingestion points: Processes user_text through the send method in SKILL.md.
Boundary markers: The skill relies on the arcjet security layer for content analysis rather than static boundary delimiters.
Capability inventory: The documentation in SKILL.md notes that the agent environment includes capabilities for bash, file system operations, and web access (web_*).
Sanitization: Implements guard_events and guard_custom_tool to validate inputs and tools against security policies.

integrate-arcjet-guard-claude-managed-agents-py