wayforth-mcp
Provides AI agents with a unified runtime to discover, use, and pay for thousands of services with automatic failover and per-run credit ceilings.
README
Wayforth
The runtime that keeps AI agents running in production.
v0.9.2
One install gives an agent a single place to discover, use, and pay for thousands of services — and when one fails mid-run, Wayforth reroutes so the agent keeps going. Agents break in production. Yours won't.
Why Wayforth
Agents demo well and fail in production. Real APIs rate-limit, expire keys, return garbage, and go down mid-task — and a long-running agent falls over the first time a dependency hiccups. Wayforth is the operational layer between your agent and that mess.
- It reroutes itself — when a service fails mid-run, Wayforth fails over to an interchangeable provider. Live today for search and model inference, expanding as coverage grows.
- It stays on budget — give any run a hard credit ceiling; the call that would exceed it is refused before it spends. No runaway loops.
- One install, every service — MCP-native, managed credentials, one balance. No per-service signups.
- Speaks the standards — MCP-native, and live A2A interop (signed agent card + JWKS + streaming).
Quick start
# 1. Install the MCP server
uvx wayforth-mcp
# Add to Claude Code permanently
claude mcp add wayforth -- uvx wayforth-mcp
# 2. Get a key at wayforth.io/signup, then:
export WAYFORTH_API_KEY=wf_live_...
Python
pip install wayforth-sdk
from wayforth import Wayforth
client = Wayforth(api_key="wf_live_...")
results = client.search("real-time stock data")
TypeScript
import { WayforthClient } from "wayforth-sdk";
const wf = new WayforthClient("wf_live_...");
Full reference: gateway.wayforth.io/guide/
What's live
- Self-heal reliability — automatic failover across interchangeable providers (search + model inference today).
- Loop budgets — per-run credit ceilings, enforced before spend, ledger-derived.
- Hosted runtime — deploy and run agents in isolated environments.
- A2A interop — the gateway serves a signed agent card and JWKS and supports streaming, verified against
a2a-sdk==0.3.26. - Managed access — a catalog of thousands of services across 19 categories, reachable through one install, with managed keys for a set of live managed services.
- Multi-rail payments (built, in sandbox) — designed to pay across card, USDC, and x402 under one balance. Live settlement is gated on incorporation.
Pricing
| Plan | Credits/month | Price |
|---|---|---|
| Free | 100 | Free |
| Starter | 6,000 | $12/mo |
| Builder | 21,000 | $29/mo |
| Pro | 72,000 | $99/mo |
| Growth | 240,000 | $299/mo |
| Enterprise | 1,000,000 | Custom |
Details: wayforth.io/pricing
On the roadmap
AP2 (verified payment authorization), live payment settlement across all rails, and expanding self-heal to more service categories.
Links
- Quickstart — wayforth.io/quickstart
- API reference — gateway.wayforth.io/guide/
- For providers — wayforth.io/providers
- PyPI (MCP) — pypi.org/project/wayforth-mcp
- PyPI (SDK) — pypi.org/project/wayforth-sdk
- Contact — wayforth.io/contact
License
Business Source License 1.1 — converts to Apache 2.0 on 2030-04-25.
© 2026 Wayforth Technologies Inc.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。