ErrorLens MCP
Provides structured error classification and recovery guidance for MCP agents, turning opaque failures into actionable retry and state information.
README
ErrorLens MCP
Structured error recovery for MCP agents.
MCP servers often fail with vague messages. ErrorLens turns those failures into structured, retry-aware, state-aware guidance that agents can use.
Why Use It
When an MCP tool call fails, agents need to know more than "something went wrong." They need to know whether retrying is safe, whether a write may have partially happened, what evidence matters, and what recovery step should come next.
ErrorLens MCP is built for that pressure point:
- Prevents blind duplicate writes after ambiguous timeouts.
- Converts opaque upstream failures into stable machine-readable categories.
- Keeps recovery guidance compact enough for agent context windows.
- Preserves successful upstream MCP responses while wrapping failures with
structured
isError: truepayloads. - Stores privacy-preserving local JSONL traces for replay and reports.
- Runs locally with no telemetry, no model dependency, and no API key.
What It Does
- Classifies opaque MCP and tool failures into a compact structured error model.
- Tells agents whether retrying is safe, whether state may have changed, and what to do next.
- Exposes a companion MCP server for diagnostics.
- Runs as a stdio MCP proxy for existing stdio upstream servers.
- Records local JSONL traces with redaction enabled by default.
- Ships a CLI for init, doctor, traces, replay, report, proxy, and adapter rule tests.
ErrorLens is a reliability layer, not a security sandbox. It does not send traces to a cloud service and it has no model or API-key dependency.
Requirements
- Node.js 22.12.0 or newer
- npm
- Windows, Linux, or macOS
Install
git clone https://github.com/Master0fFate/ErrorLens-MCP.git
cd ErrorLens-MCP
npm install
npm run build
npm run verify
Run the CLI from source:
node dist/cli/index.js --help
Or link it locally while developing:
npm link
errorlens --help
The compiled package uses portable Node APIs and is designed to work on Windows, Linux, and macOS.
Companion MCP Server
Build the project first, then add the companion server to an MCP client:
{
"mcpServers": {
"errorlens": {
"command": "node",
"args": ["/absolute/path/to/mcp-errorlens/dist/companion/diagnostic-server.js"]
}
}
}
Tools:
classify_errorrecommend_recoveryreplay_tracesummarize_failuresgenerate_adapter_rulerules_test
Proxy Mode
Create a local config:
node dist/cli/index.js init --dir .errorlens
Edit .errorlens/config.yaml, then run:
node dist/cli/index.js proxy --config .errorlens/config.yaml
For stdio upstreams, ErrorLens lists the upstream tools and forwards calls. Tool
execution failures are returned as normal MCP tool results with isError: true
and a structured ErrorLens JSON payload. Successful tool responses are preserved.
CLI Commands
node dist/cli/index.js init --dir .errorlens
node dist/cli/index.js doctor --config .errorlens/config.yaml
node dist/cli/index.js report --trace .errorlens/traces.jsonl
node dist/cli/index.js replay --trace .errorlens/traces.jsonl
node dist/cli/index.js traces --trace .errorlens/traces.jsonl
node dist/cli/index.js rules test --config .errorlens/config.yaml
node dist/cli/index.js proxy --config .errorlens/config.yaml
Demo
The repo includes a fake broken MCP server used by the QA smoke tests:
npm run verify
node dist/qa/companion-smoke.js rate-limit
node dist/qa/proxy-smoke.js write-timeout
The proxy demo shows a write-like timeout classified as SIDE_EFFECT_UNKNOWN
with retry.safe=false, which prevents blind duplicate writes.
Privacy
- No telemetry.
- Local trace files only.
- Redaction enabled by default.
- Environment variable values are never printed by
doctor. - ErrorLens improves recovery semantics; it does not provide formal security isolation.
Development
npm run lint
npm run build
npm test
npm run verify
The CI workflow runs the verification suite across Windows, Linux, and macOS.
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。