The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided specifications and requirements to generate execution plans. While it enforces a rigid markdown structure, there is a theoretical surface where malicious input requirements could be formatted into the generated plan.
Ingestion points: Processes external specifications and requirements as defined in the skill overview and metadata.
Boundary markers: Employs strict template headers and task structures, though it lacks explicit 'ignore embedded instructions' warnings for the input data.
Capability inventory: Generates Python code snippets, shell commands for testing (pytest), and version control commands (git).
Sanitization: Relies on the agent's internal logic to transform requirements into the specified plan format; no explicit sanitization of input data is defined.
[COMMAND_EXECUTION]: The skill generates specific shell commands for execution by an engineer or another agent. These commands are limited to standard development tools (pytest, git) and are essential to the skill's primary purpose of implementation planning.
[DYNAMIC_EXECUTION]: The skill generates Python code blocks and unit tests based on provided specifications. This script generation is a core functional requirement and adheres to standard templates without using unsafe evaluation methods.

implementation-planning