The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads an installation script for the Mergify CLI from the vendor's official GitHub repository (https://raw.githubusercontent.com/Mergifyio/mergify-cli/main/install.sh). It also manages the installation of several vendor-owned plugins and reporters for various test frameworks.
[REMOTE_CODE_EXECUTION]: The skill provides instructions to install the Mergify CLI by piping a remote script directly to a shell (curl | sh). This is restricted to the author's own infrastructure and is a required step for specific CI providers like CircleCI and Jenkins.
[COMMAND_EXECUTION]: The skill executes multiple commands to manage the repository and verify the setup, including GitHub CLI (gh) operations for secret management and log retrieval, and package manager commands (npm, pnpm, yarn, pip, poetry, bundle) for dependency installation.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes untrusted data from the repository's files and CI logs to determine the environment and verify success.
Ingestion points: Repository manifest files (e.g., package.json, Gemfile, pyproject.toml), lockfiles, CI configuration files, and CI job logs via gh api.
Boundary markers: No specific boundary markers or "ignore instructions" warnings are utilized for the ingested content.
Capability inventory: The skill has capabilities to write to repository files, manage CI secrets, and execute various system and package management commands.
Sanitization: The skill relies on specific grep patterns and structural parsing (like jq) to identify relevant information in logs, but does not perform comprehensive sanitization of all ingested data.

mergify-test-engine-setup