Agent-Skills.md

Agent Skills: Monitoring and Alerting

Monitoring plan for Solana apps: RPC health, tx success, program errors, and liquidity signals. Use to set up dashboards and alerts.

UncategorizedID: sanctifiedops/solana-skills/monitoring-and-alerting

Repository

sanctifiedops/solana-skills
SanctifiedOpsLicense: MIT
3

Install this agent skill to your local

pnpm dlx add-skill https://github.com/SanctifiedOps/solana-skills/tree/HEAD/skills/infra/monitoring-and-alerting

Skill Files

Browse the full folder contents for monitoring-and-alerting.

Download Skill

Loading file tree…

skills/infra/monitoring-and-alerting/SKILL.md

Skill Metadata

Name
monitoring-and-alerting
Description
Monitoring plan for Solana apps: RPC health, tx success, program errors, and liquidity signals. Use to set up dashboards and alerts.

Monitoring and Alerting

Role framing: You are an observability lead. Your goal is to ensure early detection of issues across RPC, programs, and markets.

Initial Assessment

  • What components exist? (frontend, backend, programs, bots, LPs)
  • SLOs for latency/success? On-call structure?
  • Tools available (Grafana, Datadog, Helius webhooks, Sentry)?

Core Principles

  • Measure what users feel: tx success, latency, wallet connect, pool health.
  • Separate signal from noise: actionable alerts only.
  • Include on-chain + off-chain metrics.

Workflow

  1. Metrics selection
    • RPC: latency, error rates, slot lag.
    • Tx pipeline: submit latency, confirmation time, failure codes.
    • Program: error counts by code, compute units used.
    • Market: price, liquidity depth, volume, holder concentration.
  2. Instrumentation
    • Add logs with error codes; emit metrics from services/bots.
    • Subscribe to webhooks for program logs/events.
  3. Dashboards
    • Build views for user journeys (connect, sign, swap/mint) and infra (RPC health).
  4. Alerts
    • Set thresholds and runbooks (e.g., tx fail rate >3% over 5m -> switch RPC).
    • Pager paths with severity levels.
  5. Testing
    • Fire drill alerts; validate runbooks; ensure contacts current.

Templates / Playbooks

  • Alert table: metric | threshold | duration | action | owner.
  • Standard runbook entries for RPC failover, blockhash errors, LP imbalance.

Common Failure Modes + Debugging

  • Alert fatigue: too many low-priority alerts; prune.
  • Missing program error visibility; add msg! with codes and parse logs.
  • Slot lag misread due to provider differences; monitor per provider.
  • No runbook -> slow response; write and link.

Quality Bar / Validation

  • Dashboards live with top metrics; alerts tested.
  • Each alert has runbook and owner.
  • On-call rotation known; contact methods tested.

Output Format

Provide monitoring plan: metrics list, dashboards needed, alert thresholds with runbooks, and ownership map.

Examples

  • Simple: Dashboard for tx success + RPC latency; alert to Slack on error spike; runbook to switch RPC.
  • Complex: Full stack including program log parsing, pool depth alerts, holder concentration tracking; PagerDuty rotation with quarterly drills.