@cdr-kit/vercel-ai

Vercel AI SDK adapter for cdr-kit. All 34 CDR tools as tool()s for generateText / streamText.

Hand the result to any AI SDK model — Claude, GPT-4, Gemini, anything — and the LLM autonomously discovers vaults, subscribes, decrypts, and answers from the plaintext.

Install

pnpm add @cdr-kit/vercel-ai @cdr-kit/agent ai @ai-sdk/anthropic

Quick start

import { CdrAgent } from "@cdr-kit/agent";
import { getVercelAITools } from "@cdr-kit/vercel-ai";
import { generateText } from "ai";
import { anthropic } from "@ai-sdk/anthropic";

const agent = new CdrAgent({
  privateKey: process.env.WALLET_PRIVATE_KEY as `0x${string}`,
  rpcUrl: "https://aeneid.storyrpc.io",
});

const tools = getVercelAITools(agent);

const result = await generateText({
  model: anthropic("claude-opus-4-7"),
  tools,
  prompt: "Find a recent trading-signal vault and tell me what the signal is.",
  maxSteps: 8,
});

console.log(result.text);

The LLM will call cdr_discover_vaults → cdr_get_vault_info → cdr_check_entitlement → cdr_subscribe_and_access autonomously, then summarize the decrypted JSON.

What you get

• All 34 CDR tools as Vercel AI tool()s (same shape generateText / streamText consume)
• Schema-driven: every tool emits a Zod parameters schema, so the LLM sees full param hints
• Composes with any other AI SDK tools you already have

Peer dependencies

• @cdr-kit/agent ≥ 0.7.0
• ai ≥ 6
• viem ≥ 2.21

Demos

• examples/vercel-ai-chatbot — full Next.js app: model + tools + CDR vault, end-to-end on Aeneid.

Links

• Full docs: https://cdrkit.xyz/docs/agent-kit/vercel-ai
• npm: https://www.npmjs.com/package/@cdr-kit/vercel-ai
• GitHub: https://github.com/Blockchain-Oracle/cdr-kit
• Vercel AI SDK: https://sdk.vercel.ai