1cent
Provides AI agents with SSRF-protected URL intelligence and 32 paid REST/MCP tools, plus a free catalog search, with payments in USDC on Base Mainnet.
README
1cent Web Intelligence for AI Agents
Public SSRF-protected URL intelligence API and remote MCP server. Production uses x402 v2, Base Mainnet USDC and PayAI. No account or API key. Telegram is the only administrative UI.
- MCP:
https://1cent.maxzoa.ru/mcp - Free product demo:
https://1cent.maxzoa.ru/v1/demo/pulse - Free live demo:
https://1cent.maxzoa.ru/v1/demo/live-pulse - Live catalog/prices:
https://1cent.maxzoa.ru/v1/catalog - Public trust status:
https://1cent.maxzoa.ru/status.json - Buyer guide:
https://1cent.maxzoa.ru/docs/getting-started
Production exposes 32 paid REST/MCP operations plus three free MCP tools:
catalog_search— find the correct operation and live price without a URL fetch;demo_url_pulse— inspect a fixed precomputed output sample without payment or network access.demo_live_url_pulse— run the real safe service against fixedexample.com, rate-limited.
Buyer setup starts with a no-payment diagnostic:
onecent doctor
The CLI refuses a paid call unless the buyer explicitly supplies a maximum amount, confirms Base
Mainnet and types the one-call confirmation. See examples/buyer-python and examples/buyer-node.
PostgreSQL is the runtime source for tool availability and atomic Base USDC prices. Config values are clean-install fallbacks. Paid URL work begins only after successful payment checks.
Safety contract
- public HTTP/HTTPS only; private, loopback, link-local and rebinding destinations fail closed;
- strict input schemas; unknown fields rejected;
- bounded redirects, response bodies, extracted text, concurrency and queue depth;
- payment identifier, request fingerprint and idempotent replay protection;
- UNKNOWN settlement never retried automatically;
- no seller private key, buyer key, Docker socket or JavaScript execution on the server;
- Streamable HTTP host/origin protection enabled for remote MCP.
Local quality checks
python -m venv .venv
.venv/Scripts/pip install --require-hashes -r requirements-dev.lock
.venv/Scripts/pip install --no-deps -e .
ruff check .
mypy
pytest -q
python scripts/validate_release.py
docker compose --env-file .env.example config
docker compose --env-file .env.example build
pip-audit --no-deps -r requirements.lock --progress-spinner off
Local Docker start
Copy .env.example to .env, replace placeholders, keep X402_ENVIRONMENT=testnet and
X402_NETWORK=eip155:84532, then:
docker compose up -d --build
docker compose ps
Container UID/GID: 10001:10001. No host bind-mounted writable directory is required.
This local example is intentionally testnet. Production mainnet activation requires owner approval,
a fresh PostgreSQL backup, production preflight, development bypass disabled and rollback readiness.
Documentation
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。
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。
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。
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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。