The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill dynamically generates and manages instructions for the project coordinator (AGENTS.md) and specialized subagents (.claude/agents/*.md). It assembles these instructions at runtime by interpolating user-provided context and discovered plugin metadata into predefined templates.
[COMMAND_EXECUTION]: The skill relies on local shell commands to perform its core functions, such as scanning the filesystem for installed plugins in the user's home directory (~/.claude/plugins/ and ~/.codex/plugins/) and using utilities like grep, wc, and unzip for file validation and quality control.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes and incorporates external data—including plugin descriptions and user-supplied project goals—directly into the instructions that guide future agent actions. This creates a risk where malicious input could influence the generated project guidelines.

project