The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides instructions for installing the GPC CLI tool via a remote shell script piped directly into the bash interpreter.
Evidence: curl -fsSL https://raw.githubusercontent.com/yasserstudio/gpc/main/scripts/install.sh | bash found in SKILL.md and references/github-actions.md.
Source: The script is hosted on the official GitHub repository of the tool's author.
[EXTERNAL_DOWNLOADS]: The skill recommends installing several Node.js packages and utilizing external security tools to protect the software supply chain.
Packages: @gpc-cli/cli, @gpc-cli/plugin-ci, and the socket security scanner.
Registry: Packages are sourced from the official npm registry.
[COMMAND_EXECUTION]: The utility script scripts/detect_gpc.mjs executes local shell commands to verify the environment state and tool configuration.
Method: Uses node:child_process execSync to run CLI commands.
Commands: Executes gpc --version, gpc auth status --json, and gpc config get app --json.
[PROMPT_INJECTION]: The documented AI-powered changelog generation feature introduces a surface for indirect prompt injection.
Ingestion points: Processes untrusted commit messages and git log data from the local repository (referenced in SKILL.md).
Boundary markers: No specific delimiters or boundary markers for the interpolated commit data are defined in the instructions.
Capability inventory: The skill possesses capabilities for command execution (gpc) and outbound network requests to AI API providers.
Sanitization: The procedure does not mention sanitization or escaping of the processed commit messages before they are passed to the AI model.

gpc-ci-integration