The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run extensively across multiple scripts (upgrade-check.py, risk_move.py, settings-rules.py) to execute shell commands such as git, gh (GitHub CLI), and various shell scripts for project management.
[DYNAMIC_EXECUTION]: Scripts like upgrade-check.py and old-skill-pointers.py use runpy.run_path to dynamically execute sibling Python scripts at runtime. This is used for modularizing maintenance tasks.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes various project files including plan.md, AGENTS.md, and masterplan.md to drive maintenance routines.
Ingestion points: Reads these Markdown and JSON files to identify outdated records, pieces to be converted to issues, or missing safety settings.
Boundary markers: The skill instructions explicitly require human approval before applying changes and recommend using clean Git checkpoints.
Capability inventory: Significant capabilities including file system modification (writing/deletion) and network communication via system tools.
Sanitization: Employs path traversal protection via the real_inside function in kit-leftovers.py to ensure operations are confined to the project directory.
[EXTERNAL_DOWNLOADS]: The skill performs update checks against the official gwpicard/ai-build-kit repository on GitHub and suggests using npx to fetch updates. These references target the skill's own vendor infrastructure and are used for legitimate update procedures.

maintain