Skip to content

SCENARIOS.md

Latest commit

 

History

History
194 lines (131 loc) · 6.79 KB

SCENARIOS.md

File metadata and controls

194 lines (131 loc) · 6.79 KB

Demo Scenarios

Four scenarios that demonstrate the DevOps Agent's capabilities. Each builds on the previous one, but they can also be run independently.

Scenario 1: "What's happening across my pipelines?"

The hero demo. Shows unified cross-tool pipeline visibility — the heterogeneous CI story that no single-tool vendor can replicate.

Setup

  • Meridian demo org with 1 application (Meridian Payment Platform) and 3 service components across Jenkins, GitHub Actions, and CloudBees Workflows
  • Ensure recent CI runs exist (push a small commit to trigger fresh GHA runs if needed)

Script

You: Show me what's happening across all our pipelines.

The agent runs /pipeline-overview all and:

  1. Discovers the Meridian Payment Platform application and its 3 linked components
  2. Pulls workflows and recent runs per component (GHA + CloudBees Workflows)
  3. Checks the Meridian_Jenkins_OSS controller status (version, health, queue)
  4. Pulls security findings and feature flag state
  5. Presents a unified dashboard table

Expected output:

Component GitHub Actions CloudBees Workflow Jenkins Security Flags
meridian-payment-api PASS FAIL — 3 HIGH 4 (platform)
meridian-web-portal PASS FAIL — 4 HIGH 4 (platform)
meridian-account-service — (Jenkins only) FAIL OSS controller 0 4 (platform)
  • GitHub Actions passing on both GHA-enabled components
  • CloudBees Workflows failing on all 3 service components
  • Jenkins controller: 2.541.2, healthy, 2 idle executors
  • 7 HIGH security findings (Gitleaks) across 2 components
  • 4 feature flags: instant-transfers (ON in production), dark-mode, new-dashboard, beta-payments-v2

Follow-up

You: Triage the web-portal failures.

The agent runs /triage meridian-web-portal and identifies the root cause: CloudBees Workflow ci.yaml fails at validation — the publish-test-results action requires a folder-name input but the workflow provides path. Classifies as CONFIG and recommends a specific YAML fix.

Cross-tool actions

You: File a Jira ticket for this and post to Slack.

The agent creates a structured Jira ticket in the SCRUM project with root cause, failing step, and fix recommendation, then posts a summary to the specified Slack channel.

Why it's compelling: One question gives you a cross-tool picture that would take 15 minutes of clicking through three separate dashboards. The agent spots the pattern: GitHub Actions healthy but CloudBees Workflows failing on all components — likely a common config issue.

Scenario 2: "Are we secure?"

The DevSecOps demo. Shows security posture assessment with cross-reference to CI status and feature flags.

Script

You: Give me the security posture across all our components.

The agent runs /security-scan all and:

  1. Pulls all open security issues
  2. Gets severity breakdowns per component
  3. Checks scanner coverage and SLA status
  4. Cross-references with feature flag state

Expected output:

Component HIGH Scanner Details
meridian-payment-api 3 Gitleaks Private Key (gateway.ts:19), Stripe Token (gateway.ts:7), API Key (gateway.ts:8)
meridian-web-portal 4 Gitleaks API Keys in analytics.ts:7, analytics.ts:14, dist/index.js:72, .env:1
meridian-account-service 0 — Clean
  • All 7 findings are UNREVIEWED with SLA tracking active
  • Scanner coverage: SAST active in 4 workflows, 6 workflow runs
  • SLA status: all WITHIN (due within 35 days of first detection)

Follow-up

You: File tickets for the critical findings.

The agent creates structured Jira tickets with severity, file paths, SLA dates, and remediation steps.

Why it's compelling: Combines security scanning data with CI status, feature flags, and project management in one view.

Scenario 3: "Ship the release"

The progressive delivery demo. Shows release orchestration + feature flags + security gates.

Script

You: Is the release candidate ready? Walk me through the deployment.

The agent runs /release-status and:

  1. Checks all components: CI status, security findings, flag states
  2. Identifies the release orchestration workflow
  3. Generates a go/no-go report

Expected output:

Release Readiness Report — Verdict: NOT READY

Component GHA CI CB Workflow Security Blocker
meridian-payment-api PASS FAIL 3 HIGH (SLA active) CI + Security
meridian-web-portal PASS FAIL 4 HIGH (SLA active) CI + Security
meridian-account-service — (Jenkins) FAIL 0 CI

Feature flag states per environment:

  • instant-transfers: ON in Production, default in Staging
  • dark-mode, new-dashboard, beta-payments-v2: default (code) in both

Blockers: 7 HIGH security findings, all CB Workflows failing, 3 new flags not yet configured for environments.

After resolving blockers

You: Blockers are fixed. Enable dark-mode in staging.

The agent runs /flag-rollout dark-mode enable staging:

  • Shows current state (default/code)
  • Confirms the environment (Staging)
  • Waits for confirmation
  • Enables the flag
  • Offers to post to Slack
You: Now enable it in production.
  • Agent warns: "This will affect production traffic. Confirm?"
  • Progressive rollout: Staging first, then Production

Why it's compelling: Progressive delivery end-to-end — release orchestration + feature flags + security gates, all from the IDE.

Scenario 4: "Morning standup briefing"

The full-breadth demo. Chains all skills into one synthesized output.

Script

You: Give me the morning briefing — what do I need to know?

The agent chains multiple skills:

  1. Pipeline status — /pipeline-overview all (quick summary)
  2. Security update — /security-scan all (changes since last check)
  3. CI health — /ci-health org (health score, controller status)
  4. Release status — /release-status (blockers, readiness)
  5. Feature flags — current flag states

Expected output:

A synthesized briefing with:

  • CI Health Score: 62/100
  • Jenkins controller: healthy (100%), version 2.541.2, 71 plugins, 1 update available
  • Per-tool pass rates: GitHub Actions 100%, CloudBees Workflows 0%
  • Security: 7 HIGH findings, SLA active, all UNREVIEWED
  • Release: NOT READY — 3 blockers (CI failures, security findings, unconfigured flags)
  • Recommended focus: fix CB Workflow config, triage security findings before SLA deadline

Follow-up

You: Post the briefing to the #engineering Slack channel.

Why it's compelling: Shows the agent as a true DevOps copilot that synthesizes across all tools, not a single-purpose utility.