Add periodic monitoring events: error log scan, config health check, stale PR/issue

[IYENTeam]

Summary

Move three monitoring jobs from OpenClaw cron to clawhip native periodic events, routed to #work-intake (1506217431465463929).

Background

These were previously OpenClaw cron jobs but should live in clawhip since:

• clawhip is the event ingress layer
• clawhip survives OpenClaw restarts
• keeps monitoring independent of the system being monitored

New Periodic Monitors

1. Error Log Scanner

[[monitors.logwatch]]
name = "openclaw-gateway"
path = "/tmp/openclaw/openclaw-{date}.log"
patterns = ["ERROR", "FATAL", "crash", "panic", "unhandled", "ECONNREFUSED"]
interval_secs = 900  # 15 min
channel = "1506217431465463929"  # #work-intake
mention = "<@1486621536520769547>"  # 이연
format = "alert"

[[monitors.logwatch]]
name = "clawhip"
path = "/Users/iyen/.clawhip/clawhip.log"
patterns = ["ERROR", "FATAL", "panic"]
interval_secs = 900
channel = "1506217431465463929"
mention = "<@1486621536520769547>"
format = "alert"

Behavior:

• Scan last N lines (configurable, default 200) for pattern matches
• On match: emit logwatch.{name}.alert event with matched lines
• Dedup: don't re-alert for the same error within cooldown window
• No match: silent

2. Config Change Health Check

[[monitors.filewatch]]
name = "openclaw-config"
path = "/Users/iyen/.openclaw/openclaw.json"
interval_secs = 600  # 10 min
on_change = "health_check"
health_url = "http://localhost:18789/health"
channel = "1506217431465463929"
mention = "<@1486621536520769547>"

Behavior:

• Track file mtime; on change, hit health_url
• If health check fails after config change: emit filewatch.{name}.unhealthy
• If healthy: silent

3. Stale PR/Issue Scanner

[[monitors.stale]]
interval_secs = 604800  # weekly (or cron = "0 10 * * 1")
repos = ["IYENTeam/Hent-ai", "IYENTeam/clawhip", "IYENTeam/openclaw-extensions"]
stale_days = 3
channel = "1506217431465463929"
mention = "<@1486621536520769547>"

Behavior:

• Weekly scan of open PRs and issues via GitHub API
• Filter items with no update in stale_days
• Emit stale.report event with summary table
• No stale items: emit brief "all clear" or silent

Routing

All three emit events routed to #work-intake:

[[routes]]
event = "logwatch.*"
sink = "discord"
channel = "1506217431465463929"
format = "alert"
mention = "<@1486621536520769547>"

[[routes]]
event = "filewatch.*"
sink = "discord"
channel = "1506217431465463929"
format = "alert"
mention = "<@1486621536520769547>"

[[routes]]
event = "stale.*"
sink = "discord"
channel = "1506217431465463929"
format = "alert"
mention = "<@1486621536520769547>"

Relation

• Supersedes deleted OpenClaw cron jobs (에러 로그 감시, 설정 변경 health check, stale PR/issue)
• Related: Add health monitor for OpenClaw gateway and Hermes #11 (health monitor for OpenClaw/Hermes)
• Both Add health monitor for OpenClaw gateway and Hermes #11 and this issue move monitoring responsibility from OpenClaw cron → clawhip

Acceptance Criteria

• logwatch monitor: pattern scan + dedup + alert
• filewatch monitor: mtime tracking + health check on change
• stale monitor: GitHub API scan + weekly report
• Events routed to #work-intake
• Tests for each monitor type