The Agent Skills Directory

Instructional Data Ingestion: The skill instructs the agent to read repository files such as README.md and various application source files to determine deployment requirements. This introduces a surface for indirect prompt injection where content within these files could theoretically attempt to influence the agent's behavior. However, this is a standard operational requirement for a deployment assistant, and the risk is moderated by the skill's narrow focus on providing deployment guidance.
Credential Handling Best Practices: The skill explicitly includes safety rules that prevent the agent from asking users to provide secret values directly in the chat. Instead, it directs users to store credentials in appropriate environments like Vercel project settings or local .env files. This aligns with secure configuration practices for managing sensitive information like POSTGRES_URL, GITHUB_CLIENT_SECRET, and JWE_SECRET.
Standard Tooling for Secret Generation: The instructions provide users with standard, locally-executable openssl commands to generate high-entropy secrets for session encryption. This approach ensures that secrets are generated in the user's controlled environment rather than by the AI or a remote service.

deploy-open-harness