The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the gh CLI and complex shell pipelines (using grep, sed, jq, cat, and xargs) to automate repository management tasks. These operations include creating repositories, managing labels, and configuring branch protection rules.
[DATA_EXPOSURE]: The instructions include examples for managing sensitive information, such as uploading SSH keys (~/.ssh/deploy_key) and variables from .env files to GitHub Secrets. This is documented as a primary function of the skill for CI/CD setup and is handled through the official gh secret commands.
[INDIRECT_PROMPT_INJECTION]: The skill ingest data from external GitHub resources (issues, pull requests, and comments) that can contain untrusted content.
Ingestion points: Data enters the agent context via gh issue list, gh pr view, and gh api calls in multiple files.
Boundary markers: There are no explicit instructions to the agent to treat issue/PR content as untrusted or to ignore embedded instructions.
Capability inventory: The skill can execute repository-level changes, manage secrets, and modify code via PRs.
Sanitization: No explicit sanitization or filtering of the content retrieved from the GitHub API is specified before it is processed or used in reports.
[DYNAMIC_EXECUTION]: The skill generates dynamic content, including shell scripts for bulk operations and HTML reports with embedded JavaScript for data visualization. These are used locally for repository reporting purposes.

gh-workflow