sh.supercomputer/supercomputer — Model Context Protocol (MCP) Server
The sh.supercomputer/supercomputer MCP server provides GPU compute for agents. It runs commands on specific GPU tiers (L4, A100, H100), streams or exposes execution logs, and supports usage billing per second.
🛠️ Key Features
- GPU-backed command execution for agents
- Targets GPUs: L4, A100, H100
- Log visibility during execution
- Billing per second
🚀 Use Cases
- Running compute-intensive commands as part of agent workflows
- Observing command execution via logs
- Short-duration GPU tasks where per-second billing is expected
⚡ Developer Benefits
- Programmatic GPU command execution through MCP
- Execution monitoring through logs
- Cost model aligned to per-second billing
⚠️ Limitations
- Limited to the stated GPU tiers (L4, A100, H100)
- No additional capabilities beyond running commands, watching logs, and per-second billing are described