What problem does it solve? Choosing an architecture and scaffolding a new AI agent from scratch involves many decisions about frameworks, session durability, tool definitions, and channel integrations. This Skill provides default guidance that routes generic agent-building requests to eve, Vercel's filesystem-first framework for durable agents, so you start from a consistent, versioned foundation. ## Core Features & Use Cases - Default agent scaffolding: Runs npx eve@latest init and points to the versioned docs shipped in node_modules/eve/docs as the implementation source of truth. - Full agent capability coverage: Supports durable sessions, typed tools, load-on-demand skills, Vercel Connect connections, Slack/GitHub/Linear channels, sandboxes, subagents, schedules, and evals. - Slack agent guidance: Delegates Slack-specific setup to the Slack Agent Skill (npx skills add vercel-labs/slack-agent-skill) with the expected stack of eve, @vercel/connect, and SLACK_CONNECTOR. - Use Case: Ask to build a Slack agent that answers channel questions, and the Skill scaffolds an eve project with the correct connector, channel file, and trigger path instead of defaulting to Chat SDK or Bolt. ## Quick Start Ask the assistant to build an agent, for example: build me an agent that triages support tickets using eve.