mcpField
A
APIbase — The API Hub for AI Agents

APIbase — The API Hub for AI Agents

Remote
apibase.proMedia & Design10Updated 2d ago

About APIbase — The API Hub for AI Agents

APIbase is an MCP and REST gateway for AI agents, aggregating tools from multiple services for pay-per-call use. Agents can discover and invoke tools through one endpoint, with USDC payments via x402 or MPP.

Key features

  1. 01Reach aggregated tools through one endpoint
  2. 02Offer MCP and REST interfaces
  3. 03Pay per call in USDC
  4. 04Support x402 and MPP payment rails
  5. 05Provide MCP storefronts for merchants

Use cases

  1. 01Have an agent query travel or weather services
  2. 02Call e-commerce or financial APIs as needed
  3. 03Let merchants offer products for agent purchases
  4. 04Connect API tools to an MCP client

MCP support

APIbase provides a Streamable HTTP MCP endpoint exposing aggregated API tools to MCP clients. Tool calls support per-call USDC payments through x402 on Base or MPP on Tempo.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://apibase.pro/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Packages
npmapibase-mcp-client1.0.4

Tools

Details

From the project README

APIbase.pro — The API Hub for AI Agents

One MCP + REST endpoint to 1385 tools from 400 providers. No signup, no subscription, no API key to start — pay per call in USDC (x402 on Base or MPP on Tempo).

Security Audit Deploy License: MIT MCP Registry Smithery MPPScan

Latest release: see the tag that equals package.json version; release notes in docs/releases/.


Quick Start

{
  "mcpServers": {
    "apibase": { "url": "https://apibase.pro/mcp" }
  }
}

That's the Claude Desktop config. Cursor, Windsurf, OpenAI Agents SDK, LangChain, Google ADK, CrewAI, Microsoft Copilot Studio, and plain REST: apibase.pro/connect.


For AI Agents

  • llms.txt — concise LLM context
  • ai.txt — plain-text agent discovery
  • mcp.json — MCP server metadata
  • MCP protocol versions — 2025-11-25, 2025-06-18, 2025-03-26, 2024-11-05, 2024-10-07 (Streamable HTTP on /mcp); revision 2026-07-28 is not served yet (the released SDK does not support it). Current list: protocolVersions in mcp.json
  • openapi.json — OpenAPI 3.1 spec
  • api-catalog — RFC 9727 discovery linkset
  • Roadmap — planned/shipped changes that affect what an agent can do or see

For Merchants — Integrator

AI payment for merchants: a merchant gets its own MCP storefront, and a buyer's agent gets a quote and pays in USDC (x402 on Base, MPP on Tempo) straight to the merchant wallet. Non-custodial; fee 0% during the pilot.


For paying agents: what protects your money

  • Fixed price before payment — the 402 response states the exact amount and recipient; see payments.
  • Payment identifier / idempotency — persist the id and signed payload before you send, re-send after a crash: agent guide.
  • Public receipts — https://apibase.pro/r/<id>.json, see receipts.
  • Several agents, one budget — buyers.
  • Tool definition fingerprint — discover-tools and /api/v1/tools/manifest.

Integrations & Registries

Full framework guides →

Registry listings: Smithery · Glama · MCP Registry (io.github.whiteknightonhorse/apibase) · PulseMCP · MPPScan


Self-Hosting

git clone https://github.com/whiteknightonhorse/APIbase.git && cd APIbase
cp .env.example .env   # set POSTGRES_PASSWORD, X402_PAYMENT_ADDRESS, provider keys
docker compose -f docker-compose.yml -f docker-compose.prod.yml up -d

Full prerequisites, verification steps, and config reference: docs/self-hosting.md.


Architecture

Single Node.js/TypeScript API server, PostgreSQL as the append-only financial source of truth, Redis for cache/rate-limiting only, multi-container Docker stack on one Hetzner server. Every tool call runs the same multi-stage pipeline (escrow-first, idempotent, content-moderated, fail-closed on payment, cache and rate limit; the idempotency stage fails open on Redis errors) — dual-rail payments (self-hosted x402 facilitator + MPP), error-code contract, and full container/pipeline detail (exact counts, live): docs/architecture.md.

License

MIT

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.