MCP server that lets AI agents share one NVIDIA GPU — cross-session model claims, real busy detection via NVML, protected eviction, and full VRAM visibility for Ollama + everything else holding memory
Latest release v0.3.1 — vram-mcp 0.3.1 · 14 Oct 2026
1 file for Codex and OpenCode: vram-mcp AGENTS.md — 749 tokens loaded in every session.
AGENTS.md A 749 tok These files are sushiHex/vram-mcp's own configuration — they tell Codex and OpenCode how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.