observability-debugging-contract
CommunityTrace logs metrics 一次串起来快速排障
Software Engineering#multi-agent#observability#distributed tracing#redaction#structured logging#error taxonomy#debug dump
Authorzxc1a1a1
Version1.0.0
Installs0
System Documentation
What problem does it solve?
为 AgentHub 定义从前端入口到 Gateway、Orchestrator、子 Agent、LLM Provider、Artifact 与 Tool Call 的全链路可观测性与排障契约,确保任何一次 Run 都能用统一关联 ID 和结构化数据快速定位问题原因,同时严格脱敏避免泄漏敏感信息。
Core Features & Use Cases
- Trace Context & ID 关联:规定
traceparent/tracestate的跨服务传播与traceId + requestId + runId的统一串联规则,覆盖 run/plan/step/agentTask/artifact/toolCall/llmRequest 等关联。 - 结构化日志标准:要求后端服务输出 JSON 结构化日志,规定稳定
eventdot.case 命名、基础字段集合与安全错误展示字段(errorCode、safeMessage、durationMs)。 - Span 命名与属性规范:定义推荐 span 名称与
agenthub.*前缀的自定义属性边界,避免将 prompt/token/用户隐私写入 span attributes。 - Metrics 与错误分类:给出覆盖延迟/流量/错误/饱和度的指标命名与低基数维度要求,并提供稳定错误码分域与用户可见错误策略。
- 脱敏与 Debug Dump:定义日志/trace/metrics/debug dump 的禁止与允许字段,提供可控脱敏的 debug dump 快照范围与关闭策略。
- 分进程与多 Agent 排障路径:明确 Gateway↔Orchestrator 分进程关键事件/字段,支持 2+ Agent、ordered_parallel、fallback、群聊消息关联排障。
Quick Start
Use the observability-debugging-contract skill to standardize your AgentHub run tracing, JSON logging, metrics, error codes, and redacted debug dumps so that Gateway-to-Orchestrator and multi-agent failures can be diagnosed quickly without exposing tokens or prompts.
Dependency Matrix
Required Modules
None requiredComponents
references
💻 Claude Code Installation
Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.
Please help me install this Skill: Name: observability-debugging-contract Download link: https://github.com/zxc1a1a1/Multi_Agent-AgentHub/archive/main.zip#observability-debugging-contract Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
Agent Skills Search Helper
Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.