The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell script located at a specific local path: ~/.claude/scripts/org-maintenance.sh. While the path is fixed, the skill depends on this external executable being present and functional.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and acts upon untrusted data from external repositories, creating an attack surface for indirect prompt injection.
Ingestion points: The agent is instructed to read bot_account and owner_account configurations from a CLAUDE.md file located within targeted organization repositories. It also processes PR metadata and worktree status data collected by the helper script.
Boundary markers: No specific delimiters or instructions to ignore embedded instructions within the CLAUDE.md file or repository metadata are provided.
Capability inventory: The skill possesses significant capabilities, including the ability to prune directories/worktrees and push code changes directly to GitHub branches using the gh CLI.
Sanitization: There is no evidence of sanitization or validation performed on the account identifiers or the metadata retrieved from the repositories before they are used to influence the agent's actions.

org-maintenance