The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh) and jq to perform administrative tasks on repositories. It executes commands such as gh api to fetch and update repository settings, rulesets, and branch protections. These are standard operations for an administrative setup tool and are restricted to the authorized user's credentials.
[EXTERNAL_DOWNLOADS]: The skill downloads and integrates a reusable GitHub Action workflow from Fandhe-AI/actions. It strictly mandates using a specific commit SHA (fed9c07d98367f77e5e2b63bca38843f46feee96) rather than a mutable branch reference like @main. This is a security best practice for supply chain integrity. The resource is owned by the same organization as the skill author.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from repository files such as CLAUDE.md, AGENTS.md, and other design documents to generate or update review guidelines.
Ingestion points: Reads .claude/rules/ruleset-policy.md, CLAUDE.md, and existing AGENTS.md files.
Boundary markers: The skill provides instructions on how to structure the AGENTS.md content (e.g., specific security, architecture, and reuse categories).
Capability inventory: File system read/write, network operations via gh api (GitHub).
Sanitization: None explicitly mentioned; the agent relies on its internal safety filters when processing these local files.

setup-repo-guards