hf-model-explorer-mcp

# hf-model-explorer-mcp An [MCP](https://modelcontextprotocol.io) server that lets AI agents search, compare, and right-size Hugging Face models — including estimating whether a model will actually fit in your available GPU VRAM before you download 15GB of weights and find out the hard way. Works with any MCP-compatible client: Claude, Google Antigravity, Cursor, and others. ## Why Picking a model off Hugging Face usually means checking the model card, guessing at parameter count, mentally doing GB-per-precision math, and hoping it fits your GPU. This wraps that into tools an agent can call directly, so you can ask things like: > "Find me a small instruction-tuned model I can run locally on a 12GB GPU" and get an actual, sized answer instead of a guess. ## Tools | Tool | What it does | |---|---| | `search_models` | Search HF Hub by query and task, sorted by downloads/likes/recency | | `get_model_details` | Full details on one model: params, license, tags, task | | `compare_models` | Side-by-side comparison of 2–4 models | | `check_hardware_fit` | Estimates VRAM needed at fp32/fp16/int8/int4 and whether it fits your GPU | | `find_similar_models` | Finds lighter/heavier/similar-sized alternatives for the same task | ## Install ```bash pip install hf-model-explorer-mcp ``` Or run directly without installing, via `uvx`: ```bash uvx hf-model-explorer-mcp ``` ## Configure ### Claude Desktop / Claude Code Add to your MCP config (`claude_desktop_config.json` or `.mcp.json`): ```json { "mcpServers": { "hf-model-explorer": { "command": "uvx", "args": ["hf-model-explorer-mcp"] } } } ``` ### Google Antigravity Add to `~/.gemini/antigravity/mcp_config.json` (path may vary by Antigravity version — check Settings → MCP Servers in the IDE): ```json { "mcpServers": { "hf-model-explorer": { "command": "uvx", "args": ["hf-model-explorer-mcp"] } } } ``` ## Example ``` User: I want to run an instruction-tuned 7B-ish m…