The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the gh CLI, git, npm, and npx to perform repository operations, run tests, and poll deployment statuses. These actions are aligned with the skill's stated purpose of managing the dev-to-deploy lifecycle.
[EXTERNAL_DOWNLOADS]: The GitHub Actions workflow in resources/vercel-preview.yml references several official and well-known actions (e.g., actions/checkout, gitleaks/gitleaks-action, actions/setup-node). These are standard utilities from trusted organizations and well-known services.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data, creating a potential surface for indirect prompt injection.
Ingestion points: The resolve-preview-url.js script fetches deployment statuses from the GitHub API, and the agent is instructed to analyze Playwright test logs and HTML traces.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used for these data sources.
Capability inventory: The skill has the ability to modify project code during auto-healing loops and execute shell commands via various CLI tools.
Sanitization: Data from the GitHub API is parsed as JSON, and CLI interactions utilize execFileSync rather than a shell, reducing injection risk.

live-dev-deploy