The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (MEDIUM): The manifest allows Bash(pip install *) and Bash(uv *), permitting the installation of arbitrary Python packages. This is a supply chain risk if the agent is manipulated into installing malicious dependencies.\n- [COMMAND_EXECUTION] (MEDIUM): The skill's primary function is the generation of executable Python code and system prompts (Phase 5). Dynamic code generation based on user input carries inherent risks of generating insecure or malicious logic.\n- [PROMPT_INJECTION] (LOW): The skill ingests untrusted user requirements during its 5-phase interview process. These descriptions are directly used to define agent identities and system prompts, creating a surface for indirect prompt injection that could influence the behavior of the generated agent system.

agent-architect