Swarmrails MCP
# Swarmrails MCP
Call any Bittensor subnet — text, image, video, code, TTS, forecasting, 3D assets — directly from Claude and other AI assistants. Also includes SharpSignal prediction market intelligence.
No TAO. No node setup. No subscription. Free to try.
## Tools
| Tool | Capability | Cost |
|---|---|---|
| `bittensor_text` | Conversational AI (Llama 3.3 70B) | $0.005 |
| `bittensor_translate` | Multilingual translation | $0.005 |
| `bittensor_reasoning` | Advanced reasoning (DeepSeek R1) | $0.05 |
| `bittensor_image` | Text-to-image (SDXL) | $0.075 |
| `bittensor_llm` | LLM inference (Nous Research) | $0.01 |
| `bittensor_forecast` | Financial & crypto forecasting | $0.05 |
| `bittensor_code` | Code generation | $0.01 |
| `bittensor_data` | Data analysis & synthesis | $0.005 |
| `bittensor_tts` | Text-to-speech (returns MP3) | $0.025 |
| `bittensor_scrape` | Web scraping & content extraction | $0.01 |
| `bittensor_multimodal` | Image + text reasoning (Gemini) | $0.02 |
| `bittensor_video` | Text-to-video MP4 (async) | $2.00 |
| `bittensor_3d` | Image-to-3D GLB asset (async) | $0.75 |
| `sharpsignal_predict` | Prediction market intelligence | $0.25 |
## Quick Start
Works out of the box with no configuration — free test mode is enabled by default.
### Claude Desktop
Add to `claude_desktop_config.json`:
```json
{
"mcpServers": {
"swarmrails": {
"command": "npx",
"args": ["-y", "swarmrails-mcp"]
}
}
}
```
### With real USDC payments
1. Send USDC on Base to: `0x14a129b3e3Bd154c974118299d75F14626A6157B`
2. Copy your transaction hash from [basescan.org](https://basescan.org)
3. Set the env var: `SWARMRAILS_API_KEY=myapp:0xYOUR_TX_HASH`
Each transaction hash is single-use. Payment protocol: [x402](https://x402.org) on Base.
## Payment Protocol
Swarmrails uses the **x402 protocol** — each USDC transaction on Base blockchain is a single-use API token. No accounts, no subscriptions, no API keys to manage.
```
Authorization: x402 <macaroon>:0xTRANSACTION_HASH
```
## License
MIT
TDQS
Scored across 14 tools
Most tools map to distinct modalities (image, video, code, etc.) with clear boundaries. However, `bittensor_text`, `bittensor_llm`, and `bittensor_reasoning` all handle text generation and could be confused without careful reading of their specific subnet specializations. The `sharpsignal_predict` tool is clearly distinct as the only prediction market service.
Thirteen tools follow a consistent `bittensor_<capability>` snake_case pattern that clearly indicates their function and source network. The `sharpsignal_predict` tool breaks this convention, though this is semantically justified as it uses Perplexity rather than Bittensor. All tools use descriptive, action-oriented nouns that align with their outputs.
Fourteen tools appropriately cover the breadth of Bittensor subnet offerings without excessive bloat. Each tool represents a distinct AI service (text, image, video, code, etc.) that earns its place in a comprehensive generative AI gateway. The count balances granularity with usability.
The set provides robust coverage of major generative AI modalities including text, image, video, 3D, audio, and code generation, plus data analysis and prediction markets. Minor gaps exist (e.g., no speech-to-text or image editing), but the surface covers the stated purpose of Bittensor subnet access comprehensively. The addition of prediction market intelligence adds valuable orthogonal functionality.