How do I install the Support skill?

Run npx skillmds@latest add lantisprime/support in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Support safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Support?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Support free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Support?

lantisprime (@lantisprime) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Support by lantisprime — Agent Skill

Name: Support
Author: lantisprime

Support

Use this skill during Phase 7 after a successful deployment to generate or update monitoring, logging, alerting, and observability artifacts that catch exceptions and issues in the deployed code. Produces platform-neutral scripts when no observability platform is configured, and integrates with Grafana/Datadog/CloudWatch via MCP when configured. Trigger after deploy, or when the user says "add monitoring", "observability", "alerts", "dashboards", or "what should we watch for?".

lantisprime 6856045 3.0 KB Updated Sep 17, 2026

File contents

Support (Phase 7)

Make the change observable in production.

Prerequisite

.claude/sdlc/gates/deploy-<task-slug>.md must exist and be signed.

Step 1 — Identify what to watch

From the plan's in-scope surface and the security/NFR sections of the architecture, enumerate:

New code paths → structured log events at key decision points
New endpoints or RPCs → request count, error rate, latency (p50/p95/p99)
New background jobs → run count, duration, failure rate
New external calls → timeout, failure, retry counters
New failure modes from the threat model → alerts

For each, write: what to measure, why, threshold for alert, who is paged, runbook link.

Step 2 — Produce artifacts

Write to .claude/sdlc/monitoring/<task-slug>/:

Logging changes — code deltas (in a follow-up PR, not this build's PR) or config changes
Metrics — platform-appropriate config (Prometheus rules, Datadog monitors, CloudWatch alarms)
Dashboards — JSON or TOML for the configured platform; a README if none is configured
Alerts — with routing (oncall rotation / channel), severity, auto-resolve conditions
Runbook stub — runbook.md with symptom → diagnostic → mitigation

Step 3 — Integrate (when configured)

If an observability platform is configured in .claude/sdlc/env.json:

Grafana/Datadog via MCP → propose dashboard and alert changes as pull requests or API calls; never auto-apply production alert changes without human confirmation
CloudWatch → propose IaC changes

Never silently mutate production monitoring.

Step 4 — Validate

Generate a synthetic failure in a non-production environment and confirm the alert fires.
Confirm the runbook entry matches what an on-call engineer would need at 3am.

Human gate

Summarize: what's now being watched, what's alerted, runbook status. Sign-off → .claude/sdlc/gates/support-<task-slug>.md.

What this skill must NOT do

Do not auto-apply changes to production alerting.
Do not skip a runbook entry ("we'll document later" is how silent outages happen).

References

docs/SDLC.md Support

Next step hint

After writing the gate file, pipe the next_suggestions conditions to skills/_shared/next-hint.sh and print any output:

printf '%s\n' \
  'support_gate_signed|run /docs to refresh the artifact index and traceability matrix' \
  | bash skills/_shared/next-hint.sh

Print any output verbatim. If the script outputs nothing, add nothing.

---name: supportdescription: Use this skill during Phase 7 after a successful deployment to generate or update monitoring, logging, alerting, and observability artifacts that catch exceptions and issues in the deployed code. Produces platform-neutral scripts when no observability platform is configured, and integrates with Grafana/Datadog/CloudWatch via MCP when configured. Trigger after deploy, or when the user says "add monitoring", "observability", "alerts", "dashboards", or "what should we watch for?".---# Support (Phase 7)Make the change observable in production.## Prerequisite`.claude/sdlc/gates/deploy-<task-slug>.md` must exist and be signed.## Step 1 — Identify what to watchFrom the plan's in-scope surface and the security/NFR sections of the architecture, enumerate:- New code paths → structured log events at key decision points- New endpoints or RPCs → request count, error rate, latency (p50/p95/p99)- New background jobs → run count, duration, failure rate- New external calls → timeout, failure, retry counters- New failure modes from the threat model → alertsFor each, write: what to measure, why, threshold for alert, who is paged, runbook link.## Step 2 — Produce artifactsWrite to `.claude/sdlc/monitoring/<task-slug>/`:- **Logging changes** — code deltas (in a follow-up PR, not this build's PR) or config changes- **Metrics** — platform-appropriate config (Prometheus rules, Datadog monitors, CloudWatch alarms)- **Dashboards** — JSON or TOML for the configured platform; a README if none is configured- **Alerts** — with routing (oncall rotation / channel), severity, auto-resolve conditions- **Runbook stub** — `runbook.md` with symptom → diagnostic → mitigation## Step 3 — Integrate (when configured)If an observability platform is configured in `.claude/sdlc/env.json`:- Grafana/Datadog via MCP → propose dashboard and alert changes as pull requests or API calls; never auto-apply production alert changes without human confirmation- CloudWatch → propose IaC changesNever silently mutate production monitoring.## Step 4 — Validate- Generate a synthetic failure in a non-production environment and confirm the alert fires.- Confirm the runbook entry matches what an on-call engineer would need at 3am.## Human gateSummarize: what's now being watched, what's alerted, runbook status. Sign-off → `.claude/sdlc/gates/support-<task-slug>.md`.## What this skill must NOT do- Do not auto-apply changes to production alerting.- Do not skip a runbook entry ("we'll document later" is how silent outages happen).## References- `docs/SDLC.md` Support## Next step hintAfter writing the gate file, pipe the `next_suggestions` conditions to `skills/_shared/next-hint.sh` and print any output:```bashprintf '%s\n' \ 'support_gate_signed|run /docs to refresh the artifact index and traceability matrix' \ | bash skills/_shared/next-hint.sh```Print any output verbatim. If the script outputs nothing, add nothing.

Frequently asked questions