PromptHub Batch is an AI video creation workspace: create tasks in one app and distribute video generation requests to Jimeng, Symphony, Xiaoyunque, and Runway.
Enter a prompt, add reference media, choose a model, aspect ratio, and duration, then submit tasks in batches and track queued, generating, completed, and downloadable results in one place.
With Codex, Claude, Cursor, or another agent, use the built-in MCP service to create, query, and manage tasks. The app handles account selection and queue scheduling, then returns task IDs and generated outputs.
What drains you most in AI video production is often not creativity, but these fragmented and tedious processes.
These actions don't create work. But they're most likely to interrupt inspiration and wear out patience.
One workspace to distribute and track AI video generation across multiple platforms.
PromptHub Batch connects Jimeng, Symphony, Xiaoyunque, and Runway, handling prompts, reference media, model parameters, account selection, queue execution, and result collection. Use the app directly or let an agent create tasks through MCP.
Works with the platforms you use most.
Reuse assets and common settings, submit faster, then track tasks and collect results automatically.
Reduce repeated form filling and page actions so single and continuous tasks move smoothly into execution.
Submit quickly from the page, track long-running jobs, and collect results when they finish.
Keep the browser workflow moving from efficient submission through result download and archiving.
After submission, stop reopening every page. See what is queued, generating, completed, or archived from the task view.
Designed around less form filling, faster submission, continuous execution, and automatic result collection.
The local app matches platform parameters, model constraints, and task capabilities without depending on a browser page or fixed account.
Reuse assets and common settings to reduce repeated uploads, form filling, and clicks. Submit both single and continuous tasks faster.
See queue progress, wait time, and any exceptions. No more uncertain waiting or anxiety over long tasks.
Automatically fetch videos and assets after task completion. No watching pages, no repetitive saving, no missing downloads.
Unified management of images, videos, and audio assets for faster task filling. Especially time-saving for multi-platform and multi-version iterations.
Try basic status viewing for free. Efficient submission, continuous execution, and auto download are activated per platform, so you only pay for what you use.
The app exposes a standard MCP stdio service on startup. Agents get a task ID immediately and can keep querying status and results.
PromptHub Batch now includes a local MCP service. Agents can create, query, and manage video tasks through the same queue scheduler and account strategy, without waiting for generation or filling in a local account ID.
Works with Codex, Claude Desktop, Claude Code, Cursor, MCP Inspector, and other stdio clients.
Use images, video, and audio together, with explicit @asset references in the prompt.
Selects accounts using login state, health, platform availability, concurrency, and queue load.
Restricts local directories and remote protocols, blocks sensitive, private-network, and unauthorized paths, and returns stable error codes.
{
"mcpServers": {
"prompthub-batch": {
"command": "<PromptHub Batch>/mcp-server"
}
}
}MCP starts automatically with the app. View and copy the exact entry path in Settings; development and packaged builds show their own path.
Supported models
seedance_2 · seedance_2_fast · seedance_2_5 · seedance_2_mini · seedance_2_vip
Video duration
Standard models: 4-15 sec · seedance_2_5: 4-30 sec
Settings shows MCP status, the stdio entry, client environment variables, and the allowed local asset directories.
The demo shows how the local app accepts tasks, queues them, tracks status, collects results, and lets an agent submit through MCP.
Pay attention to MCP status, the entry path, and allowed local asset directories in Settings.
The app and MCP share the same local task data, queue, and scheduler after startup.
Open the development build or packaged app to start the local task queue.
Sign in and finish local account initialization; MCP starts automatically afterward.
View MCP status, the stdio entry, client snippets, and allowed local asset directories.
Paste the configuration into Codex, Claude, Cursor, or another compatible MCP client.
Choose a PromptHub Batch license period. Local queue and MCP capabilities are included.
Monthly access
Annual access
One-time purchase, long-term access
Contact support after choosing an edition to receive the app package, license, and setup guidance.

Scan to add WeChat customer service
Working hours: 9:00-18:00
It supports Jimeng, Symphony, Xiaoyunque, and Runway, plus seedance_2, seedance_2_fast, seedance_2_5, seedance_2_mini, and seedance_2_vip.
No. MCP is built into PromptHub Batch and starts automatically after local account initialization.
No. The local queue selects accounts automatically using login state, health, platform availability, concurrency, and queue load.
An agent receives a task ID immediately without waiting for generation, then can query status, remote result URLs, and local output files.
MCP supports idempotencyKey, local path restrictions, remote request limits, and stable error codes for reliable agent retries and failure handling.
MCP only reads configured local asset directories and blocks sensitive files, unauthorized paths, localhost, and private-network addresses. Image, video, and audio references are converted for each platform.