The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local bash script (~/.claude/skills/skills-janitor/scripts/fix.sh) to perform maintenance tasks. This script handles the logic for repairing frontmatter and pruning orphaned files.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted content from various SKILL.md files in user and project directories, creating a surface for indirect prompt injection.
Ingestion points: Processes skill files within ~/.claude/skills, ./.claude/skills, and ~/.agents/skills.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore content within the files being processed.
Capability inventory: The skill uses the Bash tool with broad access to read, write, and remove files within the target directories.
Sanitization: The instruction set does not specify content sanitization; however, the skill mitigates risk by requiring a dry-run preview and user confirmation before execution.

janitor-fix