Agent Skills: v0

How to Use

Base URL: https://api.v0.dev/v1

1. Generate App or UI from Prompt

Create a new chat to generate code from a natural language prompt:

Write to /tmp/v0_request.json:

{
  "message": "Build a React dashboard with a sidebar, stats cards, and a line chart using Tailwind CSS"
}

Then run:

curl -s -X POST "https://api.v0.dev/v1/chats" --header "Authorization: Bearer $V0_TOKEN" --header "Content-Type: application/json" -d @/tmp/v0_request.json

The response includes:

• id — chat ID (used for follow-up messages)
• webUrl — link to the chat on v0.app
• latestVersion.demoUrl — live sandbox preview of the generated app
• latestVersion.files — generated source files

2. Iterate — Send Follow-up Message

Refine generated code by sending a follow-up to an existing chat. Replace <chat-id> with the ID from the previous response:

Write to /tmp/v0_request.json:

{
  "message": "Add a dark mode toggle to the header and make the sidebar collapsible"
}

Then run:

curl -s -X POST "https://api.v0.dev/v1/chats/<chat-id>/messages" --header "Authorization: Bearer $V0_TOKEN" --header "Content-Type: application/json" -d @/tmp/v0_request.json

3. Get Chat Details (Retrieve Generated Code)

Fetch the full chat including generated files. Replace <chat-id> with the target chat ID:

curl -s "https://api.v0.dev/v1/chats/<chat-id>" --header "Authorization: Bearer $V0_TOKEN"

The latestVersion.files array contains the generated source code files.

4. List All Chats

curl -s "https://api.v0.dev/v1/chats" --header "Authorization: Bearer $V0_TOKEN" | jq '.data[] | {id, name, webUrl}'

5. Create Project

Organize related chats under a named project:

Write to /tmp/v0_request.json:

{
  "name": "Customer Dashboard",
  "description": "Internal analytics dashboard for customer team",
  "privacy": "private"
}

Then run:

curl -s -X POST "https://api.v0.dev/v1/projects" --header "Authorization: Bearer $V0_TOKEN" --header "Content-Type: application/json" -d @/tmp/v0_request.json

To create a chat within a project, add "projectId": "<project-id>" to the chat creation request.

6. List Projects

curl -s "https://api.v0.dev/v1/projects" --header "Authorization: Bearer $V0_TOKEN" | jq '.data[] | {id, name, privacy}'

7. List Deployments

curl -s "https://api.v0.dev/v1/deployments" --header "Authorization: Bearer $V0_TOKEN" | jq '.data[] | {id, webUrl, status}'

8. Chat Completions (OpenAI-compatible)

Use the v0-1.0-md model directly for lightweight code generation without project context. This follows the standard OpenAI chat completions format:

Write to /tmp/v0_request.json:

{
  "model": "v0-1.0-md",
  "messages": [
    {"role": "user", "content": "Create a login form with email and password fields using shadcn/ui and Tailwind CSS"}
  ]
}

Then run:

curl -s -X POST "https://api.v0.dev/v1/chat/completions" --header "Authorization: Bearer $V0_TOKEN" --header "Content-Type: application/json" -d @/tmp/v0_request.json | jq '.choices[0].message.content'

Context window: 128,000 tokens input · 32,000 tokens output

Guidelines

1. Premium plan required: API access requires a v0 Premium or Team subscription with Vercel usage-based billing enabled
2. Iterate with follow-ups: Use POST /chats/<id>/messages to refine code — it preserves full conversation context and is more token-efficient than creating a new chat
3. demoUrl is the live preview: Each chat response contains latestVersion.demoUrl — the live sandbox URL to preview the generated app; webUrl is the link to the v0.app chat interface
4. Use projects for organization: Group related chats under a project (via projectId) to share instructions and context across conversations
5. Rate limits: 1,000 chat messages and 100 deployments per day — prefer iterating within existing chats over creating new ones