AI SDK — build AI agents, chatbots, and RAG with the Vercel AI SDK
Work the Vercel AI SDK the current way — generateText/streamText, ToolLoopAgent, AI Gateway as default provider, bundled-docs-first lookups, always-fresh model IDs, type-safe useChat consumption
- What
- Work the Vercel AI SDK the current way — generateText/streamText, ToolLoopAgent, AI Gateway as default provider, bundled-docs-first lookups, always-fresh model IDs, type-safe useChat consumption
- Cost
- Free
- Needs
- a Node.js project; the Vercel AI SDK (the skill installs only the `ai` package first — provider packages like @ai-sdk/openai and client packages like @ai-sdk/react come later, as needed)
- Install
- Copy the installer prompt below into your Muse — your agent does the rest.
Curated by Skill Harbor — @vercel-labs's ai-sdk skill (from the open-agents collection): a discipline guide for building AI features with the Vercel AI SDK without tripping over outdated APIs. The core doctrine — "do not trust internal knowledge" — forces the agent to verify everything against the bundled docs in node_modules/ai/docs/ and source before writing code, install only the ai package first (providers and client packages later, as needed), default to the Vercel AI Gateway provider when choosing models, always fetch current model IDs from the gateway endpoint (newest version wins — never from memory), run typecheck after changes, and keep options minimal. It covers building and consuming agents with the ToolLoopAgent pattern, framework-specific consumption (detect the stack from package.json first), type-safe useChat via InferAgentUIMessage, devtools setup, and a common-errors reference for renamed parameters — with the pragmatic escape hatch of ai-sdk.dev doc search for older versions. Honest caveats: deliberately paranoid — the docs-first ritual adds lookups to every task, which is the point but also the cost; assumes a Node.js project with the SDK installed, not a general AI primer; model-ID advice will age, which is exactly why it insists on fetching. License: the manifest records MIT (the manifest makes faith); the repo frontmatter carries no license field. Skill Harbor never reviews the code, review it yourself before use. Discovered via skills.sh.
Version:
Install
Prerequisites: a Node.js project; the Vercel AI SDK (the skill installs only the `ai` package first — provider packages like @ai-sdk/openai and client packages like @ai-sdk/react come later, as needed) Install "AI SDK — build AI agents, chatbots, and RAG with the Vercel AI SDK" for me. It gives my agent @vercel-labs's AI SDK discipline: never trust internal knowledge — verify APIs against node_modules/ai/docs/ and source before writing code, default to the Vercel AI Gateway provider, always fetch current model IDs from the gateway (newest wins, never from memory), build agents with the ToolLoopAgent pattern, consume them with type-safe useChat (InferAgentUIMessage), detect the project framework from package.json for framework-specific patterns, run typecheck after changes, and check the common-errors reference when typecheck fails. MIT licensed (per the discovery manifest; the repo frontmatter carries no license field). Repository: https://github.com/vercel-labs/open-agents/blob/main/.agents/skills/ai-sdk/SKILL.md 1. Fetch the SKILL.md file (and any helper files) from the repository path into a temporary folder and summarize what it does in one or two sentences. 2. Safety check: review the SKILL.md and scripts for anything suspicious (unexpected network calls, shell commands, credential harvesting). This repo should contain zero secrets in code. Verify that holds here; STOP on any red flag and tell me. 3. Install it as a skill: copy SKILL.md and its helper files into the agent's skills directory, in a folder named "ai-sdk". 4. Verify with no network calls: frontmatter valid, files in place. 5. Report what was installed, where, and what I still need to do myself (e.g. install the ai package in my Node project, choose the model/provider — AI Gateway recommended). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.
Questions
How do I install a build?
Every product page includes a copy-paste install prompt. Paste it into your Muse and it sets the build up for you — no manual configuration.
Where does my money go?
Straight to the seller. Skill Harbor never processes payments: checkout happens on the seller’s own page, usually Stripe.
What does the ✓ next to a creator’s name mean?
It means we confirmed the identity of the person behind the listing. It says nothing about the code itself — always check a build before installing it.