The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the gh (GitHub) CLI and jq to interact with repository settings. It performs administrative actions such as modifying branch protection rulesets, enabling security scanning features, and configuring merge strategies via the GitHub API (gh api).- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill incorporates external logic by adding reusable GitHub Actions workflows from the Fandhe-AI/actions repository to the target environment. It follows supply-chain security best practices by pinning these workflows to a specific, verified commit SHA (fed9c07d98367f77e5e2b63bca38843f46feee96) rather than using mutable branch references.- [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by deploying AGENTS.md, which serves as a configuration file for downstream AI review agents. * Ingestion points: The agent context is designed to read AGENTS.md and .claude/rules/. * Boundary markers: Not explicitly enforced in the template generation. * Capability inventory: The skill modifies repository workflows and uses high-privilege gh api writes. * Sanitization: The skill provides static templates for these files without dynamic sanitization mechanisms. While this creates a configuration surface, the skill instructs the user to define specific security and architectural boundaries.

setup-repo-guards