The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill functions as a documentation and template resource for software development.- [COMMAND_EXECUTION]: The skill contains configuration examples for Docker and Kubernetes, as well as Python templates for executing registered functions. These are provided as static text for educational purposes and are not executed by the agent on the host system.- [DATA_EXFILTRATION]: No hardcoded credentials, sensitive file path access, or unauthorized network operations were detected. The code templates demonstrate standard API integration patterns.- [PROMPT_INJECTION]: The instructions are consistent with the stated purpose of AI assistant development and do not attempt to override safety filters or extract system prompts.- [INDIRECT_PROMPT_INJECTION]: The provided code templates (e.g., in resources/implementation-playbook.md) demonstrate how to build LLM applications that process user input and conversation history. While this introduces a standard attack surface for LLM applications, the skill correctly provides these as architectural examples.
Ingestion points: User requirements in SKILL.md and conversation history processing in implementation-playbook.md.
Boundary markers: The templates demonstrate basic structure but do not include specific sanitization logic, which is expected for architectural guides.
Capability inventory: The code examples include simulated NLU processing, dialog management, and LLM API interactions.
Sanitization: Not present in the templates, as they are high-level design examples.

llm-application-dev-ai-assistant