The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The script scripts/scaffold_breakdown.py uses the importlib.util module to dynamically load and execute a Python script from a calculated relative path (../../add-subfeature/scripts/manage_subfeatures.py). This dynamic loading and execution of code from outside the skill's own directory creates a runtime dependency on external files that are executed within the agent's context.
[COMMAND_EXECUTION]: The skill's scaffolding functionality in scripts/scaffold_breakdown.py performs file system operations, including directory creation and file writing, based on the target argument provided at the command line. This allows the skill to modify the project structure dynamically.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes repository-level artifacts that are often controlled by various contributors.
Ingestion points: Reads various project documents including discover.md, system-design.md, ui-design.md, and user-stories.md to drive the decomposition process.
Boundary markers: Absent. The skill instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings when reading these files.
Capability inventory: The skill has the capability to write to the file system and execute dynamic Python modules through scripts/scaffold_breakdown.py.
Sanitization: Absent. The skill does not validate or escape the content gathered from the ingested markdown files before incorporating it into the final planning artifacts.

breakdown