Skip to main content
Glama
@openvecta/mcp

@openvecta/mcp

by openvecta

@openvecta/mcp

MCP server for OpenVecta — call OpenVecta inference models (OpenAI-compatible, USDC-settled on Solana) as tools from any MCP client: Claude Desktop, Cursor, Windsurf, agent kits.

npm

Tools

Tool

What it does

Auth needed

list_models

Catalog + per-1M-token USDC pricing, modality, flat fee

none (public)

estimate_cost

Estimate a call's USDC cost before running it

none (public)

chat

Chat completion on any model

keyed or x402

embed

Embeddings

keyed only

Related MCP server: @bridgenode/mcp

Two auth modes (auto-detected from env)

KEYED — set OPENVECTA_API_KEY (ov_sk_...). Uses your prepaid balance / free tier. Simplest; works everywhere. (Wins if both are set.)

X402 — set SOLANA_PRIVATE_KEY (base58). Pays per call in USDC from that wallet via the x402 protocol — no account, no key. Chat only. ⚠️ Real funds move. Test on devnet first (SOLANA_NETWORK=devnet).

Install

No install step needed — npx fetches it on demand:

npx -y @openvecta/mcp

Add to Claude Desktop / Cursor

Config file — Claude Desktop: %APPDATA%\Claude\claude_desktop_config.json (Windows) or ~/Library/Application Support/Claude/claude_desktop_config.json (macOS).

Keyed:

{
  "mcpServers": {
    "openvecta": {
      "command": "npx",
      "args": ["-y", "@openvecta/mcp"],
      "env": { "OPENVECTA_API_KEY": "ov_sk_live_..." }
    }
  }
}

x402 — pay per call, no account (test on devnet first!):

{
  "mcpServers": {
    "openvecta": {
      "command": "npx",
      "args": ["-y", "@openvecta/mcp"],
      "env": {
        "SOLANA_PRIVATE_KEY": "base58-secret-key",
        "SOLANA_NETWORK": "devnet"
      }
    }
  }
}

Restart the client, then ask e.g. "list the OpenVecta models" or "use OpenVecta's glm-5.2 to explain Solana".

Env reference

Var

Default

Notes

OPENVECTA_API_KEY

—

ov_sk_ key → KEYED mode

SOLANA_PRIVATE_KEY

—

base58 secret → X402 mode

SOLANA_NETWORK

mainnet

mainnet | devnet

SOLANA_RPC_URL

per-network

override RPC

OPENVECTA_BASE_URL

https://api.openvecta.com/v1

staging / self-host

How x402 mode works

Uses the official @x402/fetch client (@x402/svm ExactSvmScheme + a @solana/kit signer) to wrap fetch: the request hits /v1/chat/completions, gets a 402 challenge, the library builds + signs the USDC transfer from your wallet, resends with the payment header, and OpenVecta settles + serves. chat requires max_tokens here (it bounds the pre-authorization ceiling; defaults to 1024 if omitted).

The public model catalogue and per-token pricing are readable without auth at /.well-known/x402.

Security: never commit a private key. For production/agents, load it from a secrets manager, not a config file. list_models / estimate_cost need no auth.

Build from source

npm install
npm run build   # -> dist/index.js

License

MIT

Available Tools

3 tools
chatA

Run a chat completion on an OpenVecta model (OpenAI-compatible). Pays via your configured mode (prepaid key or x402 per-call). Returns the reply + token usage.

Inputs: model (required), messages (required), max_tokens, temperature.

embedA

Create embeddings for text with an OpenVecta embeddings model (e.g. 'text-embedding-3-small'). Keyed mode only (x402 is chat-only).

Inputs: input (required), model (required).

estimate_costA

Estimate the USDC cost of a call before running it: input/output token counts priced against the model's live rate (+ flat per-call fee).

Inputs: model (required), input_tokens (required), output_tokens.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.9/5.0

    Scored across 3 tools

    Disambiguation5/5

    Each tool serves a distinct purpose: chat for text generation, embed for text embeddings, and estimate_cost for cost estimation. No overlap in functionality.

    Naming Consistency4/5

    Names are descriptive verbs or verb-noun combos, but 'chat' and 'embed' are single words while 'estimate_cost' uses an underscore, a minor inconsistency.

    Tool Count4/5

    Three tools is minimal but appropriate for the focused scope of running AI model operations and cost estimation. The count is on the low side but not insufficient.

    Completeness4/5

    Covers the core workflows: chat completions, embeddings, and cost estimation. Missing model listing or management, but the surface is complete for basic usage.

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      F
      maintenance
      Enables MCP clients to access all endpoints of an x402 gateway by paying real-time microtransactions (USDC on Base) per API call, with automatic tool discovery and spend guardrails.
      23
      64 npm
      MIT
    • A
      license
      A
      quality
      C
      maintenance
      Enables AI agents to make pay-as-you-go AI inference requests through x402 with automatic Solana USDC payments, no API keys or registration, while enforcing configurable spending limits.
      3
      89 npm
      1
      MIT No Attribution
    • F
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to access 11 paid x402 endpoints as standard MCP tools, paying per call in USDC on Base without API keys, covering chat, code, vision, embeddings, crypto prices, weather, geolocation, forex, and WHOIS data.
      -
    • A
      license
      A
      quality
      C
      maintenance
      Enables MCP-capable AI agents to make pay-per-call Solana data requests (snapshots, risk checks, token reports, health, balances, transactions, trending pairs) with automatic USDC settlement over x402.
      5
      MIT