AgentSkillsCN

monitor

搭建可观测性体系——包括结构化日志、健康探针、指标监控以及告警机制（马库斯·陈的工作流程）

SKILL.md
--- frontmatter
name: monitor
description: Set up observability — structured logging, health probes, metrics, and alerting (Marcus Chen's workflow)
disable-model-invocation: true

Set up monitoring and observability: $ARGUMENTS

1. Structured Logging Audit

Verify the current logging setup:

  • • Consistent log format across all modules
  • • Level: INFO by default, DEBUG when debug mode enabled
  • • No print() statements in source root
  • • All loggers use logging.getLogger(__name__)
  • • Project's audit logger used for state changes

Log levels used correctly:

LevelUsed For
DEBUGVerbose troubleshooting (protocol commands, request details)
INFONormal operations (startup, shutdown, state changes)
WARNINGRecoverable issues (resource not found, degraded mode)
ERRORFailures that need attention (communication errors)

2. Health Endpoint Verification

Verify the health endpoint returns appropriate status information:

  • • Responds within 200ms (load balancer requirement)
  • • Bypasses authentication
  • • Returns degraded status when external resource is disconnected
  • • Returns correct version from config

3. Readiness Probe

Consider adding a readiness endpoint that checks:

  • •Service is initialized
  • •External resource is connected and responsive
  • •Rate limiter is functional

4. Request Logging Middleware

Consider adding request/response logging for production debugging:

  • •Log method, path, status code, and response time for every request
  • •Don't log request/response bodies (may contain sensitive data)
  • •Use a dedicated access logger

5. Metrics Endpoint (optional)

If Prometheus/metrics are needed:

  • •Request count by endpoint and status code
  • •Request latency histogram
  • •External resource connection state gauge
  • •Active entity states gauge

6. Docker Health Integration

Verify Dockerfile HEALTHCHECK matches the health endpoint URL (see project config).

Output

For each section, report:

  • •CONFIGURED: Already in place and working
  • •ADDED: New monitoring capability implemented
  • •RECOMMENDED: Optional improvement (describe what and why)