The Agent Skills Directory

External Dependency Management: The skill provides instructions for installing the @vercel/connect package using standard Node.js package managers. This is an expected procedure for integrating Vercel Connect functionality into a project.
Command Line Interaction: The skill leverages the vercel connect CLI to create and manage connections. These operations are performed within the context of the user's authenticated Vercel environment, which is typical for cloud service management tools.
Sensitive Data Handling: Instructions are provided for managing OAuth tokens, including recommendations to use shell variables to prevent accidental exposure in logs. The skill references environment variables like VERCEL_OIDC_TOKEN for authentication, which is a standard practice for secret management.
Indirect Prompt Injection Surface: The skill provides a mechanism for an agent to interact with external services (such as Slack or GitHub) based on user instructions. This creates a potential surface where external content or user-supplied data could influence agent behavior if not properly handled by the underlying implementation.
Ingestion points: User prompts requesting interactions with external services (SKILL.md).
Boundary markers: The skill does not explicitly define delimiters for untrusted data in the provided documentation.
Capability inventory: The skill uses vercel connect CLI commands and performs network requests via HTTP APIs or the SDK (SKILL.md).
Sanitization: Explicit sanitization or filtering of external content is not detailed in the skill instructions.

vercel-connect