The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's architecture is designed to ingest and process data from external, untrusted sources, creating a surface for indirect prompt injection.
Ingestion points: The skill utilizes createWebhook and createHook to receive external payloads (references/api-deep-dive.md) and accepts user messages for AI agent workflows (SKILL.md).
Boundary markers: The provided code examples do not demonstrate the use of delimiters or specific instructions to the agent to disregard embedded commands in external data.
Capability inventory: The skill utilizes the fetch tool for network operations (SKILL.md, references/api-deep-dive.md) and supports arbitrary business logic and I/O within step functions.
Sanitization: The documentation lacks examples of input validation, schema enforcement, or sanitization of external payloads before they are used to influence workflow state or agent behavior.
[EXTERNAL_DOWNLOADS]: The documentation encourages the installation of multiple Node.js packages from public registries, including workflow, @workflow/ai, and several database-specific backend implementations.
[COMMAND_EXECUTION]: The skill provides CLI commands for development, management, and observability, such as npx workflow web and npx workflow inspect, which are executed by the user.

vercel-workflow