The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/verify_extraction.py invokes the git show command using the subprocess.run function. This is used to read previous versions of a file from the repository history for comparison. The implementation uses a list of arguments, which is a secure method for executing external processes.
[PROMPT_INJECTION]: The skill processes content from repository Markdown files, creating an indirect prompt injection surface.
Ingestion points: Content is read from CLAUDE.md and .claude/rules/*.md files via scripts/audit_claude_md.py and scripts/verify_extraction.py.
Boundary markers: No explicit boundary markers or instruction-ignoring warnings are used when processing the text.
Capability inventory: The skill possesses capabilities to read and write files, as well as execute git commands.
Sanitization: The scripts do not perform semantic sanitization, focusing instead on structural analysis and verbatim comparison.
[SAFE]: The skill accesses and reads CLAUDE.md and related .claude/rules/ files. This access is limited to the repository context and is necessary for the skill's primary function of managing memory configuration files.

slim-claude-md