MCPEvent
An MCP server that provides a test suite with regression testing for event management, featuring isolated subprocess per test file and hard timeouts to prevent hangs.
README
MCP Event Server — Test Suite
Feature-focused, fast, and hang-safe regression tests for the MCP event server
(server.py, events.py, store.py, runtime.py, sources/).
Layout
test/
run_all.py # regression driver: isolated subprocess per file, hard 300s/file timeout
helpers/ # lifecycle, mcp, wait, mock_http, runner — shared test support
mcp_result.py # SDK result normalization + safe teardown
test_*.py # one feature per file (the units of the suite)
config.json # base server config (overwritten per-test, restored after)
TEST_RUNTIME_MAP.md # per-file server usage, harness behavior, run guidance
Production source under test/ is frozen — do not edit server.py, events.py,
store.py, runtime.py, errors.py, client.py, config.json, requirements.txt,
or sources/* as part of test work.
Run
# Focused (preferred while debugging one feature):
python test/test_acknowledgement.py
python test/run_all.py --group fast
# Full regression (CI / release only — slow path):
python test/run_all.py
python test/run_all.py --timeout 300
See TEST_RUNTIME_MAP.md for the server-usage matrix, the bounded-wait strategy, and
why you should run focused, not full, during development.
Principles
- No server unless the test needs the MCP/HTTP boundary. Application logic (store, events, routing, checkpoint/ack, background tasks, source lifecycle, dedup) is tested directly against the real objects with an in-memory stub bus.
- One server per D-level file (exceptions: restart tests in
test_errors.py/test_reconnect.py, andtest_subscriptions.pywhose tests need different source configs). - Bounded waits, no long fixed sleeps. Readiness/counts are polled with timeouts.
- Hard timeout (300 s/file) + parent-owned cleanup — the runner terminates the child's
process group, so a hung file fails fast (reported as TIMEOUT) and any
server.pyit spawned is killed with it. No reliance on the child'satexit(which can't run after a hard kill).
Deleted (do not recreate)
test/test_phase8.py, test/integrate_test.py — replaced by the per-feature files above.
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。