What problem does it solve? LLMs often carry outdated knowledge about the Vercel platform, such as recommending Edge Functions, assuming streaming requires the edge runtime, or referencing deprecated products like Vercel Postgres and Vercel KV. This Skill injects authoritative corrections at session start so generated code and architecture advice reflect the current platform. ## Core Features & Use Cases - Knowledge Corrections: Overrides stale assumptions about Edge Functions, streaming/SSE, Fluid Compute, Node.js 24 defaults, 300s timeouts, Active CPU pricing, and 5 GB function package sizes. - New Product Awareness: Introduces recent Vercel offerings including Eve, AI Gateway, Queues, Sandbox, Rolling Releases, BotID, and the Vercel MCP server. - vercel.ts Configuration: Promotes the typed vercel.ts config (via @vercel/config) as the replacement for vercel.json, with examples for rewrites, redirects, headers, and crons. - Use Case: When a user asks how to stream an AI response on Vercel, the assistant correctly recommends the default Node.js Fluid Compute runtime instead of unnecessarily switching to runtime = 'edge'. ## Quick Start Ask the assistant to configure a Vercel project or answer a Vercel platform question, and the injected knowledge updates will automatically correct any outdated assumptions.
