The Agent Skills Directory

[PROMPT_INJECTION]: The skill operates on untrusted user specifications and incorporates them into implementation plans that include direct commands for the agent (e.g., 'REQUIRED SUB-SKILL'). Ingestion points: The skill processes user-provided 'spec or requirements' as defined in the overview. Boundary markers: Absent; the skill does not define delimiters or instructions to ignore embedded directives in the processed input. Capability inventory: The skill can write files to the 'docs/plans/' directory, execute shell commands (git, pytest), and dispatch subagents via 'superpowers:subagent-driven-development'. Sanitization: Absent; user input is interpolated directly into implementation steps and code blocks without validation or escaping.
[COMMAND_EXECUTION]: The skill generates detailed plans containing shell commands for git operations and test execution (pytest), and it encourages the creation of Python scripts that are subsequently executed in the local environment.

writing-plans