VAP Media · Unified MCP Server for AI Agents (Flux · Veo · Suno)
Execution control layer for AI agents - Reserve, execute, burn/refund pattern for media generation
README
VAP Media · Unified MCP Server for AI Agents (Flux · Veo · Suno)
Generate AI images, videos, and music with deterministic pricing.
Get Started
No setup required – start generating in 2 minutes:
- Create Agent → Get your API key instantly
- Add Funds → Start from $1
- Generate → Start creating images, videos, and music
Why VAP?
VAP is an MCP server that enables image, video, and music generation directly from agentic workflows. It exposes generative media capabilities as portable, pay-per-use tools usable across Claude Desktop, Cursor, and other MCP-compatible clients.
When AI agents work with paid APIs, they need:
- Cost visibility – Know exactly what you'll pay before execution
- Retry control – Bounded, predictable retry behavior
- Clear ownership – Every task tracked and accountable
- Enterprise auth – OAuth 2.1 M2M for secure integrations
VAP provides this control layer.
What VAP Does
VAP is an MCP Server that provides Flux image generation, Veo 3.1 video generation, and Suno V5 music generation with full cost control.
Supported AI Models:
- Flux2 Pro – Photorealistic images
- Veo 3.1 – Cinematic videos
- Suno V5 – Original music
Production Pipeline:
- FFmpeg post-processing (format conversion, audio normalization)
- Automatic quality optimization for broadcast standards
- Permanent cloud storage with instant CDN delivery
How it works:
- Pre-commit pricing – Know exact cost before execution
- Reserve → Burn → Refund – Hard budget guarantees
- Deterministic behavior – Predictable results every time
- Explicit ownership – Every task has a clear owner
How It Works
Agent: "Generate an image of a sunset"
↓
VAP: "Reserving cost..."
VAP: "Reserved. Executing..."
VAP: "Success. Here's your image."
If it fails:
Agent: "Generate an image of a sunset"
↓
VAP: "Reserving cost..."
VAP: "Reserved. Executing..."
VAP: "Failed. Full refund. Error: Provider timeout"
Your agent never sees the complexity. It just gets deterministic results.
Presets
| Type | Preset |
|---|---|
| Image | image.basic |
| Video | video.basic |
| Music | music.basic |
| Campaign+SEO | streaming_campaign |
| Full Production+SEO | full_production |
All media productions are automatically normalized and delivered through a fast, orchestrated pipeline in accordance with defined broadcast quality standards.
Pricing: See vapagent.com for current pricing.
MCP Integration
Step 1: Get Your API Key
Option A: Dashboard (Recommended)
Go to vapagent.com/dashboard/signup.html and create your agent.
Option B: API
curl -X POST https://api.vapagent.com/v3/agents \
-H "Content-Type: application/json" \
-d '{"name": "my-agent"}'
Save the api_key from the response. It's shown only once.
Step 2: Activate Your Account
Deposit minimum $1 to unlock generation capabilities:
curl -X POST https://api.vapagent.com/v3/deposits/init \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{"amount": 5.00, "provider": "crypto"}'
Step 3: Configure Your MCP Client
Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": {
"vap": {
"url": "https://api.vapagent.com/mcp",
"transport": "streamable-http",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}
Cursor
Add to .cursor/mcp.json (project) or ~/.cursor/mcp.json (global):
{
"mcpServers": {
"vap": {
"command": "npx",
"args": [
"mcp-remote",
"https://api.vapagent.com/mcp",
"--header",
"Authorization: Bearer YOUR_API_KEY"
]
}
}
}
Cline (VS Code)
Add to Cline MCP settings:
{
"mcpServers": {
"vap": {
"command": "python",
"args": ["path/to/mcp/vap_mcp_proxy.py"],
"env": {
"VAP_API_KEY": "your_api_key"
}
}
}
}
Restart your client after configuration.
Available Tools
| Tool | What it does |
|---|---|
generate_image |
Create photorealistic image from text (Flux2 Pro) |
generate_video |
Create cinematic video from text (Veo 3.1) |
generate_music |
Create original music from text (Suno V5) |
estimate_cost |
Preview cost before generating |
check_balance |
Check your current balance |
get_task |
Check status of a running task |
list_tasks |
List your recent tasks |
Alternative: Local Proxy
For environments that don't support headers, use the local proxy:
{
"mcpServers": {
"vap": {
"command": "python",
"args": ["/path/to/mcp/vap_mcp_proxy.py"],
"env": {
"VAP_API_KEY": "your_api_key"
}
}
}
}
MCP Registry: io.github.elestirelbilinc-sketch/vap-e
OAuth 2.1 (Enterprise)
For enterprise integrations, VAP supports OAuth 2.1 M2M (machine-to-machine) authentication via Scalekit.
How It Works
OAuth Token → MCP Proxy → Validate → Resolve client_id → Agent → Execute
Setup
# 1. Get OAuth token from your identity provider
curl -X POST "https://your-tenant.scalekit.dev/oauth/token" \
-d "grant_type=client_credentials" \
-d "client_id=m2m_xxx" \
-d "client_secret=your_secret"
# 2. Link OAuth client to your VAP agent (one-time)
curl -X PUT "https://api.vapagent.com/v3/agents/me/oauth" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"oauth_client_id": "m2m_xxx"}'
# 3. Use OAuth token with MCP
# The MCP proxy validates tokens and maps to linked agents
Benefits
- SSO Integration – Connect VAP to your existing identity provider
- No API Key in Config – Tokens rotate automatically
- Audit Trail – OAuth events logged separately
SDK Usage
Installation
pip install vap-sdk
Image Generation
from vape_client import VAPEClient
client = VAPEClient(api_key="your_api_key")
result = client.generate(
description="A serene mountain landscape at sunset"
)
print(f"Image URL: {result.image_url}")
print(f"Cost: ${result.cost}")
Video Generation
# Generate video with Veo 3.1
video = client.generate_video(
prompt="Cinematic aerial shot of coastal cliffs at golden hour",
duration=8,
aspect_ratio="16:9",
generate_audio=True
)
print(f"Task ID: {video.task_id}")
# Poll for completion
task = client.get_task(video.task_id)
print(f"Status: {task.status}")
print(f"Video URL: {task.result_url}")
Music Generation
# Generate music with Suno V5
music = client.generate_music(
prompt="Upbeat indie folk song with acoustic guitar and warm vocals",
duration=120,
instrumental=False
)
print(f"Task ID: {music.task_id}")
# Check task status
task = client.get_task(music.task_id)
print(f"Audio URL: {task.result_url}")
Async Usage
import asyncio
from vape_client import AsyncVAPEClient
async def main():
async with AsyncVAPEClient(api_key="your_api_key") as client:
# Image
image = await client.generate(description="A futuristic cityscape")
print(f"Image URL: {image.image_url}")
# Video
video = await client.generate_video(prompt="Ocean waves at sunset")
print(f"Video Task: {video.task_id}")
# Music
music = await client.generate_music(prompt="Lo-fi chill beats")
print(f"Music Task: {music.task_id}")
asyncio.run(main())
Task Management
# List recent tasks
tasks = client.list_tasks(limit=10)
for task in tasks.tasks:
print(f"{task.task_id}: {task.status} - {task.task_type}")
# Get specific task
task = client.get_task("your-task-id")
print(f"Result: {task.result_url}")
API Endpoints
| Endpoint | Method | Description |
|---|---|---|
/v3/generate |
POST | Create media execution task |
/v3/tasks/{id} |
GET | Retrieve task status |
/v3/tasks/{id}/result |
GET | Retrieve task result |
/v3/balance |
GET | Check account balance |
/v3/agents/me/oauth |
PUT | Link OAuth client (Enterprise) |
/v3/agents/me/oauth |
GET | Check OAuth link status |
Full API Docs: api.vapagent.com/docs
The Four Guarantees
1. Pre-Commit Pricing
Every task has a known cost before execution. No surprises.
2. Budget Enforcement
Set a max budget. VAP enforces it. Hit the limit? Task rejected. Balance protected.
3. Failure Ownership
Every task has an explicit owner. Every failure has an address. No more "the agent did something and I don't know what."
4. Deterministic Production Quality
Every output is normalized to broadcast standards. Consistent formats, predictable quality, publish-ready media. No variance between runs.
Links
- MCP Registry: registry.modelcontextprotocol.io
- API Documentation: api.vapagent.com/docs
- MCP Endpoint:
https://api.vapagent.com/mcp
License
MIT License – see the LICENSE file for details.
VAP Media · Unified MCP Server for AI Agents
"Deterministic media production for AI agents."
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器