# AgentCompute

> AgentCompute is a free MCP server for design and media. Shared GPU compute beta with an account API, task catalog and MCP access. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Free
- **Platforms:** Web, API
- **Website:** https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/agents
- **MCP Registry name:** io.github.Ockram4444/agentcompute
- **Categories:** design and media, AI models and LLM tooling, cloud and infrastructure
- **Listing updated:** 2026-11-15

## About

AgentCompute pools GPU capacity from people who choose to share it, and lets agents run jobs on that pool. The registry entry lists text, image and audio work, and the site describes a published service catalog that is machine-readable.

There are two sides. To use compute, you can try a small anonymous demo in Studio, then sign in for continuing use, job history and persistent API and MCP credentials. To provide compute, you install the client, sign in with Google, finish setup and choose Start sharing; jobs then run during sharing without per-job approvals.

It is a free beta. Signed-in accounts have no small daily cap, but waits depend on compatible providers being available, with fair scheduling and bounded queues. The site asks you to review results and submit only authorized, non-confidential content.

## Features

- **Anonymous demo:** A small trial in Studio needs no sign-in.
- **Persistent credentials:** Signed-in accounts get job history and lasting API and MCP credentials.
- **Machine-readable catalog:** Publishes its available services as a catalog agents can read.
- **Contribute your GPU:** Providers install a client and start sharing capacity.
- **Fair scheduling:** Uses bounded queues and fair scheduling when capacity is tight.

## Good for

- Run text, image or audio jobs from an agent without owning a GPU
- Try a small demo before signing in
- Offer spare GPU time to others
- Read the service catalog programmatically to choose a task

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http agentcompute https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/v1/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/v1/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "agentcompute": {
      "url": "https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/v1/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "agentcompute": {
      "type": "http",
      "url": "https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/v1/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "agentcompute": {
      "serverUrl": "https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/v1/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "agentcompute": {
      "type": "http",
      "url": "https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/v1/mcp"
    }
  }
}
```

## Tools

- `get_capability_catalog`: Read the common published-service catalog, accepted output types, account requirement, submission/result tools and per-service live status paths. This is not a live availability claim. Experimental candidates are excluded. Check status before submission.
- `submit_catalog_job`: Experimental structured extraction from one to six nonconfidential product records supplied as text. No URL fetching or OCR. Check catalog_extraction.available first. One task unit, bounded retries; retain idempotency_key. Output includes JSON/CSV with source quotes and null for missing data. Schema and source matching are checked, semantic correctness is not guaranteed. Poll get_studio_job.
- `submit_agent_job`: Experimental local worker agent. Version 1 reads supplied documents. Set agent_task.version=2 for decimal calculations, CSV preview/filter/numeric sort/summary/group sums, and TXT/Markdown/CSV/JSON files using approved local CPU/RAM functions. Check services.agent_tools for version 2. Up to eight tool calls, eight optional documents, 24000 input characters and three files (16000 bytes each, 24000 
- `get_compute_status`: Check prepared audio-provider and queue capacity. Languages and the exact profile are returned; audio availability is separate from Studio. Account daily limits are absent; anonymous-demo eligibility is reported separately. Returns availability, permitted task types, and limits. Does not start a job.
- `submit_compute_job`: Beta users must supply their uploaded asset_id and kind compute_job. No accuracy guarantee is made for uploaded files. Owner legacy calls can queue the fixed synthetic benchmark. compute_job returns a transcript; delegated_task additionally requires word error rate at most 10% against the known reference. Returns a job_id to poll with get_compute_job. No payment is charged. Reuse the same idempote
- `get_compute_job`: Read a previously submitted job by job_id. Returns status, transcript, runtime and reference quality when complete. For queued or running jobs, wait poll_after_seconds before checking again. Failed or expired jobs are not successful results.
- `get_service_status`: Public live availability, model catalog, separate account and anonymous-demo limits and polling guidance. No token required. Check before submitting; an offline GPU cannot execute work.
- `submit_chat_job`: Default routing fairly rotates compatible idle providers. Optional parallel:{parts:2|3,objective,documents} splits document analysis and returns one synthesized result under the same job ID. Each part and synthesis count toward quotas. Otherwise provide messages. Routing fastest requires comparable configured benchmarks; lowest_cost uses configured prices, with fair fallback when metrics are unkno
- `submit_image_job`: Queue a free text-to-image task from a complete scene description. Each task is independent, not an edit of a previous image. Returns a job ID; poll get_studio_job then get_studio_image for the PNG. Configured model and size appear in service status/results. No external AI fallback.
- `get_studio_job`: Read your own chat, image, agent, catalog or coordinated-analysis job. Includes partial text while running, final text or image availability, GPU runtime and server timings. Wait poll_after_seconds between checks; stop on succeeded, failed or expired. Does not create or retry work.
- `get_studio_image`: Return the completed image as an MCP image content block. Requires the same token/identity that submitted the job. Images expire after 24 hours from submission; download before expiry.

## FAQ

### Is it free?

It is described as a shared free beta.

### Is there a daily cap?

The site says no small daily cap for signed-in accounts, though waits depend on provider availability.

### What should I submit?

Only authorized, non-confidential content, and review the results.

### How do I share my GPU?

Install the client, sign in with Google, complete setup and choose Start sharing.

### Is AgentCompute free?

Yes. AgentCompute is free to use.

### How do I add AgentCompute to Claude Code?

Run this command in your terminal: claude mcp add --transport http agentcompute https://u8x5gpk0ul.execute-api.eu-north-1.amazonaws.com/v1/mcp

### What are the best alternatives to AgentCompute?

The closest alternatives to AgentCompute are HyperFrames by HeyGen, ffmpeg-skill, Claude Code Ultimate Guide, AvatarSay — Fictional Character & Talking Video MCP and ChartAnt.

## Alternatives

- [HyperFrames by HeyGen](https://ai-powered.com/mcp/hyperframes-by-heygen): HyperFrames by HeyGen is a freemium MCP server for design and media.
- [ffmpeg-skill](https://ai-powered.com/mcp/ffmpeg-skill): ffmpeg-skill is an open-source MCP server for design and media.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [AvatarSay — Fictional Character & Talking Video MCP](https://ai-powered.com/mcp/avatarsay-fictional-character-and-talking-video-mcp): AvatarSay — Fictional Character & Talking Video MCP is a freemium MCP server for design and media.
- [ChartAnt](https://ai-powered.com/mcp/chartant): ChartAnt is a freemium MCP server for design and media.
- [VfxSmith](https://ai-powered.com/mcp/vfxsmith): VfxSmith is a freemium MCP server for design and media.
- [CalderFlow](https://ai-powered.com/mcp/calderflow): CalderFlow is a free MCP server for design and media.
- [Eikino](https://ai-powered.com/mcp/eikino): Eikino is a freemium MCP server for security.

---
Source: https://ai-powered.com/mcp/agentcompute
