{
  "markdown": "# @voxell/forge-mcp\n\nAn MCP server for **Forge** — Voxell's hosted text-embedding API. It exposes Forge to any\nMCP client (Claude, Cursor, Cline, Windsurf, VS Code, …) as two tools:\n\n- **`embed`** — turn text into vectors\n- **`list_models`** — list available models and their dimensions\n\nYou bring a Forge API key. The server is stateless, and **Voxell does not store the text you\nsend or the vectors it returns** — only usage metadata (token counts) is recorded, for billing.\nIt does embeddings only — no storage, no search, no RAG. Those are different products.\n\n## Quick install\n\nOne-click install in your editor (then replace `your-key-here` with a real key from\n[dash.voxell.ai](https://dash.voxell.ai)):\n\n[![Add to Cursor](https://cursor.com/deeplink/mcp-install-dark.svg)](cursor://anysphere.cursor-deeplink/mcp/install?name=forge&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkB2b3hlbGwvZm9yZ2UtbWNwIl0sImVudiI6eyJGT1JHRV9BUElfS0VZIjoieW91ci1rZXktaGVyZSJ9fQ==)\n[![Install in VS Code](https://img.shields.io/badge/VS_Code-Install-0098FF?style=flat-square&logo=visualstudiocode&logoColor=white)](vscode:mcp/install?%7B%22name%22%3A%22forge%22%2C%22command%22%3A%22npx%22%2C%22args%22%3A%5B%22-y%22%2C%22%40voxell%2Fforge-mcp%22%5D%2C%22env%22%3A%7B%22FORGE_API_KEY%22%3A%22your-key-here%22%7D%7D)\n\n**Claude Code** — one command:\n\n```bash\nclaude mcp add forge -e FORGE_API_KEY=your-key-here -- npx -y @voxell/forge-mcp\n```\n\nAny other client (Claude Desktop, Cline, Windsurf, Zed, …) uses the standard `mcpServers`\nblock — see [Use it](#use-it) below.\n\n## Why Forge\n\n- **Quality you can dial.** Forge runs the Qwen3-Embedding family; `ultra` is the 8B — ~75+\n  average task score on MTEB, currently #4 on MTEB (English), and the top *usable* model (the\n  three ranked above it are research-only). `turbo` (0.6B) is the fast/cheap default. Pick your\n  quality/cost point.\n- **Matryoshka (MRL).** Set `dim` to truncate (re-normalized) for ~4× smaller, cheaper vectors.\n- **Low latency** (Go + CUDA engine), **zero-trust** (per-key auth; mTLS available), and **free to\n  start** (10M tokens, no card — [dash.voxell.ai](https://dash.voxell.ai); more at\n  [voxell.ai/forge](https://voxell.ai/forge)).\n\n## What you can do with it\n\n- **Add semantic search** — embed your documents with `input_type: \"document\"` and each query\n  with `input_type: \"query\"`, then rank by cosine similarity.\n- **Build RAG** — embed a knowledge base, store the vectors, and retrieve the closest chunks to\n  ground an LLM.\n- **Find similar or duplicate text** — embed two texts and compare their vectors.\n- **Cluster or classify** — embed a batch, then cluster or train a classifier on the vectors.\n- **Shrink vector storage** — set `dim` to truncate (Matryoshka) and trade a little accuracy\n  for smaller, cheaper vectors.\n- **Straight from your editor** — ask your AI agent (Cursor, Claude, …) to embed a snippet, a\n  batch, or a file via the `embed` tool — no separate script.\n\n## Requirements\n\n- Node.js ≥ 18 (tested on 20)\n- A Forge API key — create one at https://dash.voxell.ai. New accounts start with 10M free\n  tokens, no credit card.\n\n## Use it\n\nMost MCP clients run it on demand with `npx`. Add this to your client's MCP config:\n\n```json\n{\n  \"mcpServers\": {\n    \"forge\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"@voxell/forge-mcp\"],\n      \"env\": { \"FORGE_API_KEY\": \"your-key-here\" }\n    }\n  }\n}\n```\n\n(Cursor, Claude Desktop, Cline, Windsurf, and VS Code all use this `mcpServers` shape.)\n\n## Tools\n\n### `embed`\n\n| arg | type | default | notes |\n|-----|------|---------|-------|\n| `input` | string or string[] | — | text(s) to embed (required) |\n| `model` | string | `turbo` | `turbo` (1024-d), `pro` (2560-d), `ultra` (4096-d) |\n| `dim` | number | model default | truncate to N dimensions (Matryoshka) — works on every model |\n| `input_type` | `\"query\"` \\| `\"document\"` | `document` | use `query` for search queries |\n\nReturns the vectors plus the model, dimension, and token count.\n\nDefault is `turbo` — the one you probably want. `pro`/`ultra` trade size and speed for more\ndimensions.\n\n### `list_models`\n\nLists the available models and their dimensions.\n\n## Configuration\n\n| env | required | default |\n|-----|----------|---------|\n| `FORGE_API_KEY` | yes | — |\n| `FORGE_BASE_URL` | no | `https://api.voxell.ai` |\n\n## Beyond MCP: OpenAI-compatible API\n\nForge speaks the **OpenAI embeddings API**. Point any OpenAI client at Forge — **no code change**,\nand your existing vector dimensions are preserved:\n\n```python\nfrom openai import OpenAI\n\nclient = OpenAI(base_url=\"https://api.voxell.ai/v1\", api_key=\"your-forge-key\")\n# the exact call you already make — now on a higher-ranked engine:\nclient.embeddings.create(model=\"text-embedding-3-large\", input=[\"hello world\"])  # -> 3072-d\n```\n\nYour OpenAI model names map to a **matching-dimension** Forge tier (`text-embedding-3-small`/\n`ada-002` → 1536-d, `text-embedding-3-large` → 3072-d), so existing vector stores slot in\nunchanged. Or address Forge tiers directly — `turbo` | `pro` | `ultra`. Also supports `dimensions`\n(Matryoshka, re-normalized) and `encoding_format: \"base64\"`.\n\n**It's an upgrade on every path.** Forge's *smallest* tier (`turbo`, Qwen3-Embedding-0.6B)\noutranks OpenAI's *largest* embedding model (`text-embedding-3-large`) on MTEB — so there's no\ndrop-in that lands worse. `ultra` (Qwen3-Embedding-8B, ~75+ average task score, #4 on MTEB English)\nis a different league.\n\n**Why re-embedding onto Forge is worth it.** Embedding is a one-way door: whatever an encoder\ndiscards at write time is gone — no reranker, longer prompt, or bigger LLM downstream reconstructs\nwhat the vectors never captured. The model you embed with sets the ceiling on everything above it.\nRe-embed once onto a higher-ranked engine and that ceiling rises — permanently.\n\n## License\n\nMIT © Voxell, Inc.\n",
  "bytes": 5842,
  "sha": "afa47cabe51925b2749abb8ebb6e7652447f721e6b3ab7e19ba9221ad611f5e3",
  "repo_slug": "voxellinc/forge-mcp",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_ai_voxell_forge_bc4d4ad9/readme"
}