The Agent Skills Directory

[SAFE]: The skill functions as an interactive workflow architect. It operates by querying the user for task requirements and then translating those requirements into project-specific configuration files such as CLAUDE.md and .claude/settings.json. These actions are performed transparently and are central to the skill's documented functionality.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it transforms user-provided task descriptions into persistent agent instructions. Ingestion points include user descriptions and existing project rules; boundary markers are established through an interactive step-by-step workflow; capabilities include writing configuration files; and no specific sanitization is performed. This surface is inherent to the skill's primary purpose as a workflow builder.
[NO_CODE]: The skill contains only instructional markdown content. It does not package any executable code, shell scripts, or external dependencies, relying entirely on the agent's native ability to follow the provided guidance.

create-harness