The Agent Skills Directory

[COMMAND_EXECUTION]: The helper script scripts/archive_todo.py executes the git command via subprocess.run to resolve the repository root. This is a common and safe practice for repository-based tooling, utilizing fixed arguments and avoiding shell execution.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled data within the .ai/TODO.md file.
Ingestion points: Content is read from .ai/TODO.md in the repository root.
Boundary markers: There are no explicit delimiters to isolate user task content from agent instructions.
Capability inventory: The associated script can perform file writes, create directories, and run git commands.
Sanitization: The script parses Markdown structure using regex but preserves the text content of tasks and prose verbatim without additional sanitization.

todo-archive