The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user input and codebase content to generate GitHub issues, creating a surface for potential instruction injection. * Ingestion points: User responses to clarifying questions and data retrieved from mckinsey/agents-at-scale-ark and mckinsey/agents-at-scale-marketplace repositories. * Boundary markers: Employs a quoted shell heredoc ('EOF') when invoking the gh CLI to prevent shell command injection from the generated issue body. * Capability inventory: Accesses codebase files and interacts with GitHub APIs through the gh tool. * Sanitization: Incorporates a mandatory human-in-the-loop verification step, requiring the user to approve the final draft before the issue is created.

issue-creation