运维监控

Name: 运维监控
Author: kanyun-inc

AI 智能已审计 v0.2.0

作者仓库星标 11
许可证 MIT
作者更新于实时读取
作者仓库 octo-cli

流狐档案事实层信息：领域、Agent、信任分、运行环境、权限与网络

领域: AI 智能 · octopus · observability · logs
兼容 Agent: Claude Code
Cursor
Cline
Codex
Windsurf
Gemini CLI

+20
信任分: 100 / 100 · 已通过审计
作者 / 版本 / 许可: @kanyun-inc · v0.2.0 · MIT
Token 消耗评级: 低消耗
接入复杂程度: 需手动接入
是否需要外部 API Key: 不需要
兼容的系统: Docker
底层运行要求: Node.js · Python · Docker
文件与系统权限: 只读
允许写入 / 修改
读取环境变量
网络行为: 仅限本地
安装命令数: 26 条

档案由构建时根据 SKILL.md 与安装命令自动衍生，可能与作者实际意图存在差异。

输出预览 octo.preview

---
name: octo
description: Query Octopus observability platform — logs, alerts, traces, metrics, issues, services, LLM, RUM…
category: AI 智能
runtime: Node.js / Python / Docker
---

# octo 输出预览

## PART A: 任务判断
- 适用问题：提示词、Agent 工作流、模型评估或自动化推理。
- 输入要求：目标材料、限制条件、期望输出和验收方式。
- 证据边界：围绕“Onboarding (first time in a project) / Step 1: Check auth / Step 2: Init (generates template + installs skill)”读取原文规则，不把推断写成作者承诺。

## PART B: 执行结果
- **01** 任务判断：确认你的需求是否属于提示词、Agent 工作流、模型评估或自动化推理，并标出输入、限制和预期结果。
- **02** 执行计划：优先按“Onboarding (first time in a project) / Step 1: Check auth / Step 2: Init (generates template + installs skill)”拆成步骤，说明每一步会读取什么、修改什么、产出什么。
- **03** 交付结果：给出可复制的命令、文件改动、检查清单或内容草稿，并说明如何继续迭代。
- **04** 风险边界：结合 读取文件、写入/修改文件、读取环境变量、主要在本地完成、通常不需要额外 API Key 给出执行前确认项。

## Running Rules
- 读取文件、写入/修改文件、读取环境变量；主要在本地完成；通常不需要额外 API Key。
- 先小样例验证，再放大到真实任务。
- 交付时同时给结果、检查口径和下一步迭代建议。

SKILL.md SKILL.md

---
name: octo
description: Query Octopus observability platform — logs, alerts, traces, metrics, issues, services, LLM, RUM…
category: AI 智能
source: kanyun-inc/octo-cli
---

# octo

## 什么时候使用
- 把 AI / Agent方向的常用动作沉淀成 Agent 可调用的技能 适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免 Agent 泛泛回答。 围绕 octopus、o…
- 面向提示词、Agent 工作流、模型评估或自动化推理，优先处理能明确输入、步骤和验收标准的工作。

## 需要提供什么
- 目标材料、目录范围、期望结果和不可改动内容。
- 是否允许联网、执行命令、读写文件或调用外部服务。

## 执行规则
- 围绕「Onboarding (first time in a project) / Step 1: Check auth / Step 2: Init (generates template + installs skill)」组织步骤，不把推断写成作者事实。
- 读取文件、写入/修改文件、读取环境变量；主要在本地完成；通常不需要额外 API Key。
- 先跑小样例，确认结果可检查后再扩大任务范围。

## 输出要求
- 给出最终产物、关键证据、验证方式和下一步动作。
- 信息不足时标记 unknown，不编造命令、平台或依赖。

证据边界与执行链路

作者原文负责流程事实；仓库文件负责来源和命令；流狐只补充适用场景、限制和质量判断。

skill "octo" {
  输入层     -> 用户目标 + 目标文件 + 禁止范围 + 验收标准
  上下文层   -> Onboarding (first time in a project) / Step 1: Check auth / Step 2: Init (generates template + installs skill)
  规则层     -> SKILL.md 触发条件 / 执行顺序 / 输出格式
  运行层     -> Node.js / Python / Docker | 读取文件、写入/修改文件、读取环境变量 | 主要在本地完成
  安全层     -> 通常不需要额外 API Key + 小任务验证 + diff / 日志复核
  输出层     -> 可复制结果 + 检查清单 + 下一步迭代
}

解读按原文结构重写，命令、链接、术语均保留；右侧可核对作者原始 SKILL.md

先判断是否适合

核心用途：把 AI / Agent方向的常用动作沉淀成 Agent 可调用的技能适合处理AI Agent、提示词、模型评估与自动化推理，核心价值是把输入、判断、执行、验证和交付边界固定下来，避免 Agent 泛泛回答。围绕 octopus、observabili…
适用任务：更适合 AI Agent、提示词、模型评估与自动化推理，尤其是输入、步骤、验收口径可以反复复用的场景。
不宜硬套：如果任务只有一次性问答、来源材料很少，或需要作者没有说明的命令和外部服务，应先降级为普通对话。

作者设计意图

触发依据：使用时机、上下文收集、输出格式和失败处理共同决定它是否适合当前任务。
来源边界：作者原文负责事实，仓库文件负责命令和文件证据；Fluxly 只补充用途、限制和使用判断。

作者的方法与取舍

执行主线：任务围绕「Onboarding (first time in a project) / Step 1: Check auth / Step 2: Init (generates template + installs skill)」组织，把触发条件、必要上下文、执行规则和交付标准连成一条可复核的链路。
设计重点：重点不在重复教程，而在解释作者为什么规定这些步骤、哪些条件不可省略，以及结果偏离时应回到哪一层检查。
使用判断：先确认任务与原文边界一致，再决定是否启用；来源没有说明的命令、平台和依赖保持未知，不由 Fluxly 补写。

边界和复核

依赖条件：通常不需要额外 API Key，适合先用小任务验证触发是否稳定。
权限提醒：会涉及 read / write / env-read 权限，执行前要说明可读写范围、命令边界和回退方式。
质量标准：结果应包含明确产物、必要证据和下一步动作；如果只是泛泛解释，需要补充输入和验收条件后重跑。

Octopus CLI — Query Observability Data

CLI tool octo-cli for querying the Octopus observability platform (octopus.zhenguanyu.com). Covers logs, alerts, error tracking, traces, metrics, services, LLM, RUM, and events.

Onboarding (first time in a project)

When the user says anything like "接入 Octopus", "set up observability", "octo init", or when you need observability data but the project has no context yet, run the full onboarding:

Step 1: Check auth

cat ~/.octo-cli/config.json 2>/dev/null || echo "NOT CONFIGURED"

Has app_id/app_secret: proceed to step 2
NOT CONFIGURED: ask the user for their Octopus ApplicationKey (appId + appSecret), then:
```
npx octo-cli login --app-id <APP_ID> --app-secret <APP_SECRET>
```

Step 2: Init (generates template + installs skill)

npx octo-cli init

This creates .claude/rules/octopus-observability.md and installs the octo skill.

Step 3: Fill in the context (YOU do this, not the user)

Read the generated template, then scan the codebase and query live Octopus data:

3a. Identify project type and find service names from code:

Scan: package.json, pom.xml, docker-compose, k8s manifests
Spring: application.yml → spring.application.name
Frontend: @octopus-sdk/browser-rum init → applicationName
Node.js: env vars SERVICE_NAME, OCTOPUS_SERVICE
Monorepo: check each workspace member independently

3b. Verify services and query live data:

# Verify services exist
npx octo-cli services list -e online -l 1d
npx octo-cli services list -e test -l 1d

# For each service: topology (real upstream/downstream)
npx octo-cli services topo <SERVICE> -e online -l 1d

# Entry points (HTTP routes, RPC methods, MQ consumers)
npx octo-cli services entries <SERVICE> -e online -l 1d

# Sample traces (see DB, caches, MQ, external calls)
npx octo-cli trace search -q "service = <SERVICE>" -e online -l 1h -n 10

# Check RUM (frontend services)
npx octo-cli rum list -q "application.name = <SERVICE>" -e online -l 1d -n 1

# Check existing issues
npx octo-cli issues search -q "service = <SERVICE>" --status unresolved -l 7d

3c. Write findings into the template, replacing all  sections.

The whole onboarding should take one conversation. After it's done, every future agent session in this project will auto-load the context.

Keeping Context Fresh

The context file is a living document, not a one-time snapshot. When you're querying Octopus during any task, watch for signs that the context is stale:

Unknown service: query returns data for a service not listed in the context file
Topology changed: services topo shows new upstream/downstream not documented
New data collection: find SDK imports (RUM, LLM, etc.) in code that aren't in the context
Service gone: a documented service returns no data in recent queries
Known issues resolved or new ones appeared

When you spot a mismatch, update the context file immediately — don't just use the correct data for your current task and leave the file stale. Future agents (and future you) depend on it being accurate.

Quick refresh commands:

# Compare documented services vs what's actually reporting
npx octo-cli services list -e online -l 1d
npx octo-cli services list -e test -l 1d

# Re-check topology for a known service
npx octo-cli services topo <SERVICE> -e online -l 1d

# Check current top issues (may have changed)
npx octo-cli issues search -q "service = <SERVICE>" --status unresolved -l 7d

Octopus URL → octo-cli 命令

When the user pastes an Octopus page URL, parse the hash parameters and convert to the equivalent octo-cli command. URL parameters are semantic and self-explanatory:

https://octopus.zhenguanyu.com/#/rum-explorer?env=test&rumExplorerQuery=...&rumExplorerQueryEventType=resource&rumExplorerSelectedApplication=rush-app&time=1d
→ npx octo-cli rum list -e test -q "<decoded rumExplorerQuery>" -l 1d

https://octopus.zhenguanyu.com/#/log-explorer?env=online&query=service%3Dmyapp&time=15m
→ npx octo-cli logs search -e online -q "service=myapp" -l 15m

https://octopus.zhenguanyu.com/#/trace-explorer?env=test&query=...&time=1h
→ npx octo-cli trace search -e test -q "<decoded query>" -l 1h

https://octopus.zhenguanyu.com/#/dashboard/detail?id=3117&env=test&time=1d&var-cluster=...
→ Dashboard — extract metric names from dashboard config, query with octo-cli metrics

https://octopus.zhenguanyu.com/#/alert?env=online&status=firing
→ npx octo-cli alerts search -e online -s firing

Key URL parameter mappings: env → -e, time → -l, query/rumExplorerQuery → -q, rumExplorerQueryEventType → determines rum type filter, rumExplorerSelectedApplication → application.name in query.

Query Syntax

All -q/--query options use Octopus search syntax:

Field Search

field = value           # equals (case-sensitive)
field != value          # not equals
field > 100             # greater than (analysis fields only)
field >= 100            # greater or equal
field < 100             # less than
field <= 100            # less or equal
field in (a, b, c)     # in list
field not in (a, b)    # not in list
field = web*            # wildcard match (case-sensitive)

Full-text Search

keyword                 # contains keyword (case-sensitive)
"exact phrase"          # exact phrase match

Logical Operators

service = myapp AND level = ERROR       # AND (default in basic mode)
level = ERROR OR level = WARN           # OR
NOT "timeout"                           # NOT
(level = ERROR OR level = WARN) AND service = myapp   # parentheses

Priority: NOT > AND > OR. Operators are case-insensitive.

Reserved Fields

Field	Description
`env`	Environment: online, test
`service`	Service name
`host`	Server hostname
`level`	Log level: FATAL, ERROR, WARN, INFO, DEBUG, TRACE
`trace_id`	Trace ID for correlation
`issue_id`	Error tracking issue ID
`k8s.pod.name`	Kubernetes pod name
`source`	Log source (nginx, docker, python, etc.)

Commands

Time Range Options (all query commands)

-l, --last <duration>    # Relative: 15m, 1h, 2d, 1w
--from <time>            # Absolute: epoch ms or ISO string
--to <time>              # Absolute: epoch ms or ISO string
-e, --env <env>          # Environment: online (default), test
-o, --output <fmt>       # Output: json (default), table, jsonl

Logs

# Search logs
npx octo-cli logs search -q "level = ERROR" -l 15m
npx octo-cli logs search -q "service = myapp AND level = ERROR" -l 1h -n 100
npx octo-cli logs search -q "service = myapp" -e test --last 30m

# Aggregate logs (count, group by)
npx octo-cli logs aggregate -q "level = ERROR" -g service -l 1h
npx octo-cli logs aggregate -q "level = ERROR" -g service:10 -a "*:count"
npx octo-cli logs aggregate -g level -l 30m

Alerts

# Search alerts
npx octo-cli alerts search -s firing -l 1h                    # firing alerts
npx octo-cli alerts search -s firing -p P0,P1 -l 6h           # P0/P1 only
npx octo-cli alerts search --service myapp -s all -l 1d        # by service

# Alert rules
npx octo-cli alerts rules --group-id -1                        # all rules
npx octo-cli alerts rules --search "error" --page 1 --page-size 10

# Silence
npx octo-cli alerts silence --rule-id 123 --alert-id 456 --duration 2h

Error Tracking (Issues)

# Search issues
npx octo-cli issues search --status unresolved -l 1h
npx octo-cli issues search --status all --sort logCount -l 1d
npx octo-cli issues search -q "service = myapp" --status unresolved

# Detail
npx octo-cli issues detail <issueId>

# Manage
npx octo-cli issues assign --user 123 --ids id1,id2
npx octo-cli issues update --ids id1,id2 -s resolved

Traces

# Search spans
npx octo-cli trace search -q "service = myapp" -l 15m
npx octo-cli trace search -q "status = error" -l 1h -n 100

# Aggregate spans
npx octo-cli trace aggregate -a "duration:p95" -g service -l 1h
npx octo-cli trace aggregate -a "*:count" -g "name" -l 30m

Metrics

# Timeseries query
npx octo-cli metrics query "sum(http_requests{service=myapp}.as_count)" -l 1h
npx octo-cli metrics query "avg(cpu_usage{service=myapp})" --points 50 -l 2h

# Single point value
npx octo-cli metrics point "sum(error_count{service=myapp}.as_count)"

Services / APM

npx octo-cli services list -l 1h                       # active services
npx octo-cli services entries myapp -l 1h               # entry points
npx octo-cli services topo myapp -l 1h                  # topology graph

LLM Observability

npx octo-cli llm -l 1h                                        # all LLM spans
npx octo-cli llm -q "model.name = gpt-4" -l 1d                # by model
npx octo-cli llm -q "application.name = myapp" -l 1h -n 50    # by app

RUM (Real User Monitoring)

npx octo-cli rum list -e test -q "application.name = rush-app AND type = session" -l 1d
npx octo-cli rum list -q "type = error" -l 1h
npx octo-cli rum detail <id>

Events

npx octo-cli events -l 1d                              # recent events
npx octo-cli events -q "type = deployment" -l 7d       # deployments

Users

npx octo-cli users alice bob                            # search by name

Typical Workflows

Investigate an Error

# 1. Find firing alerts
npx octo-cli alerts search -s firing -p P0,P1

# 2. Search ERROR logs for the affected service
npx octo-cli logs search -q "service = affected-service AND level = ERROR" -l 30m

# 3. Check error tracking issues
npx octo-cli issues search -q "service = affected-service" --status unresolved

# 4. Find error traces — these show the FULL call chain across services
npx octo-cli trace search -q "service = affected-service AND status = error" -l 30m

# 5. If a trace has a trace_id, search logs across ALL services in that trace
npx octo-cli logs search -q "trace_id = <TRACE_ID>" -l 1h

# 6. Check service topology to understand blast radius
npx octo-cli services topo affected-service -l 1h

Cross-Service Debugging (Trace-Driven)

# Start from a trace_id (from logs, alerts, or user reports)
# 1. Find all spans in this trace across all services
npx octo-cli trace search -q "trace_id = <TRACE_ID>" -l 1d -n 100

# 2. Search logs from ALL services involved in this trace
npx octo-cli logs search -q "trace_id = <TRACE_ID>" -l 1d

# 3. If the root cause is in a downstream service, check its topology
npx octo-cli services topo downstream-service -l 1h

# 4. Check if the downstream has its own error patterns
npx octo-cli issues search -q "service = downstream-service" --status unresolved

Map Service Dependencies

# 1. Get the full topology graph
npx octo-cli services topo myapp -e online -l 1d

# 2. List entry points (HTTP routes, RPC methods, MQ consumers)
npx octo-cli services entries myapp -e online -l 1d

# 3. Sample recent traces to see call patterns and latency
npx octo-cli trace search -q "service = myapp" -l 1h -n 20

# 4. Check P95 latency per entry point
npx octo-cli trace aggregate -a "duration:p95" -g name -q "service = myapp" -l 1h

Understand Service Health

# 1. Error rate by service
npx octo-cli logs aggregate -q "level = ERROR" -g service:10 -l 1h

# 2. P95 latency by service
npx octo-cli trace aggregate -a "duration:p95" -g service -l 1h

# 3. Service topology
npx octo-cli services topo myapp -l 1h

# 4. Recent deployments (may correlate with issues)
npx octo-cli events -q "service = myapp" -l 1d

Monitor LLM Usage

# 1. List recent LLM calls
npx octo-cli llm -l 1h -n 20

# 2. Filter by model
npx octo-cli llm -q "model.name = claude*" -l 1d

Tips

Default environment is online. Use -e test for test environment.
Default time range is 15 minutes. Always specify -l for broader searches.
Use -o jsonl for piping to jq or wc -l.
Logs search max 500 per request; use --scroll-id from response for pagination.
Query syntax is case-sensitive for field names and values, case-insensitive for operators.
Wildcards (*) only work in field search, not full-text search.
Use double quotes for exact phrase matching or special characters in queries.

Deep-Dive Skills

This repo bundles specialized Octopus skills for deeper domain knowledge. Install individually as needed:

# Install all at once
npx reskill@latest install github:kanyun-inc/octo-cli/skills/octopus-log-query -a claude-code cursor -y
npx reskill@latest install github:kanyun-inc/octo-cli/skills/octopus-metrics -a claude-code cursor -y
npx reskill@latest install github:kanyun-inc/octo-cli/skills/octopus-rum -a claude-code cursor -y
npx reskill@latest install github:kanyun-inc/octo-cli/skills/octopus-llm-trace -a claude-code cursor -y
npx reskill@latest install github:kanyun-inc/octo-cli/skills/octopus-data-collection -a claude-code cursor -y
npx reskill@latest install github:kanyun-inc/octo-cli/skills/octopus-openapi -a claude-code cursor -y
npx reskill@latest install github:kanyun-inc/octo-cli/skills/octopus-web-sdk-helper -a claude-code cursor -y

Skill	Focus
`octopus-log-query`	Log search syntax, charting, log-to-metric, tokenization
`octopus-metrics`	Metric types (Count/Gauge/Histogram), QL syntax, as_count/as_rate, expressions
`octopus-rum`	RUM concepts (Session/View/Action/Error), Web SDK, Core Web Vitals
`octopus-llm-trace`	LLM Trace SDK (Java/TS/Python), span kinds, cost tracking
`octopus-data-collection`	Log/Trace/Metric collection (HTTP, Kafka, javaagent, Node.js, Python)
`octopus-openapi`	OpenAPI signing (V1/V2), SDK integration, all HTTP endpoints
`octopus-web-sdk-helper`	Web SDK troubleshooting, config guidance, sourcemap upload

安装命令

一键复制安装选择包管理器,点击右侧复制,粘贴到终端回车即可

npx skills add kanyun-inc/octo-cli/skills -y

查看通用安装教程

查看 Skill 原文

兼容工具

Claude Code
Cursor
Cline
Codex
Windsurf
Gemini
OpenCode
Roo Code
AAide
+Augment
GitHub Copilot
OpenClaw
KKiro
Junie
WWarp
Goose
Trae
Kilo Code
OpenHands
Amp
Zencoder
CContinue
AIAider
CodeBuddy
CodeGeeX
Replit Agent

安全审计

100 / 100

已审计

所有静态检查均通过，未发现问题。

元信息

来源: github
kanyun-inc/octo-cli
版本: 0.2.0
许可证: MIT
English

分类与标签

分类 AI 智能

标签