launch

launch

Purpose

Orchestrate the full deployment and marketing pipeline. Launch coordinates testing, deployment, live site verification, marketing asset creation, and public announcement. One command to go from "code ready" to "product live and marketed."

- ALL tests must pass before any deploy attempt. Zero exceptions. Block deploy if any of: tests failing, TypeScript errors present, build fails, or sentinel CRITICAL issues detected.

Triggers

• /rune launch — manual invocation
• Called by team when delegating launch tasks

Calls (outbound)

• test (L2): pre-deployment full test suite
• audit (L2): pre-launch health check — full 7-phase quality gate
• deploy (L2): push to target platform
• incident (L2): if post-launch health check fails → triage and contain
• retro (L2): post-launch retrospective — what went well, what didn't
• browser-pilot (L3): verify live site screenshots and performance
• marketing (L2): create launch assets (landing copy, social, SEO)
• watchdog (L3): setup post-deploy monitoring
• video-creator (L3): create launch/demo video content
• L4 extension packs: domain-specific launch patterns when context matches (e.g., @rune/devops for infrastructure, @rune/ecommerce for storefront)

Called By (inbound)

• User: /rune launch direct invocation
• team (L1): when team delegates launch phase

---

Execution

Step 0 — Artifact Readiness Check

Before starting the pipeline, verify that prerequisite artifacts exist. Scan using Glob — do NOT hardcode paths, use discovery patterns.

REQUIRED ARTIFACTS:
  Source code:        Glob **/*.{ts,tsx,js,jsx,py,rs,go} — at least 1 match
  Build config:       Glob {package.json,Cargo.toml,pyproject.toml,go.mod} — at least 1 match
  Tests:              Glob **/*.{test,spec}.* OR **/test_*.* — at least 1 match

RECOMMENDED ARTIFACTS (warn if missing, don't block):
  Design system:      Glob .rune/design-system.md — if frontend project
  Deploy config:      Glob {vercel.json,netlify.toml,Dockerfile,fly.toml,.github/workflows/*} — any 1
  README:             Glob README.md
  Environment:        Glob .env.example OR .env.production — warn about secrets if .env found

BLOCKING CONDITIONS:
  ❌ No source code found → STOP: "Nothing to deploy"
  ❌ No build config found → STOP: "No project config detected — cannot determine build/deploy"
  ❌ No tests found → WARN: "No tests detected — pre-flight will run build-only verification"

Report artifact status before proceeding:

## Artifact Check
- Source: ✅ [N] files ([language])
- Build config: ✅ [file]
- Tests: ✅ [N] test files | ⚠️ No tests found
- Deploy config: ✅ [platform] | ⚠️ Not found (will detect in Phase 2)
- Design system: ✅ .rune/design-system.md | ⚠️ Not found (run /rune design first for UI projects)

Step 1 — Initialize TodoWrite

TodoWrite([
  { content: "PRE-FLIGHT: Run full test suite and verification", status: "pending", activeForm: "Running pre-flight checks" },
  { content: "DEPLOY: Detect platform and push to production", status: "pending", activeForm: "Deploying to production" },
  { content: "VERIFY LIVE: Check live URL and setup monitoring", status: "pending", activeForm: "Verifying live deployment" },
  { content: "MARKET: Generate landing copy and social assets", status: "pending", activeForm: "Generating marketing assets" },
  { content: "ANNOUNCE: Present all marketing assets to user", status: "pending", activeForm: "Preparing announcement" }
])

---

Phase 1 — PRE-FLIGHT

Mark todo[0] in_progress.

REQUIRED SUB-SKILL: rune:verification
→ Invoke `verification` with scope: "full".
→ verification runs: type check, lint, unit tests, integration tests, build.
→ Capture: passed count, failed count, coverage %, build output.

Block deploy if ANY of: [ ] Tests failing (failed count > 0) [ ] TypeScript errors present [ ] Build fails [ ] sentinel CRITICAL issues detected (invoke rune:sentinel if not already run)

If any check fails: → STOP immediately → Report: "PRE-FLIGHT FAILED — deploy blocked" → List all failures with file + line references → Do NOT proceed to Phase 2

Mark todo[0] completed only when ALL checks pass.

---

Phase 2 — DEPLOY

Mark todo[1] in_progress.

2a. Detect deployment platform.

Bash: ls package.json
Read: package.json  (check "scripts" for deploy, build, start commands)

Platform detection (in order):
  1. Check package.json scripts for "vercel" → platform = Vercel
  2. Check package.json scripts for "netlify" → platform = Netlify
  3. Check for vercel.json or .vercel/ dir → platform = Vercel
  4. Check for netlify.toml → platform = Netlify
  5. Check for Dockerfile or fly.toml → platform = custom/fly.io
  6. Fallback: ask user for deploy command before continuing

2b. Execute deploy command.

Vercel:
  Bash: npx vercel --prod
  Capture: deployment URL from stdout

Netlify:
  Bash: npx netlify deploy --prod --dir=[build_output_dir]
  Capture: deployment URL from stdout

Custom (package.json script):
  Bash: npm run deploy
  Capture: deployment URL or status from stdout

Fly.io:
  Bash: flyctl deploy
  Capture: deployment URL from stdout

Error recovery:
  If deploy command exits non-zero:
    → Capture full stderr
    → Report: "DEPLOY FAILED: [error summary]"
    → Do NOT proceed to Phase 3
    → Present raw error to user for diagnosis

Mark todo[1] completed when deploy returns a live URL.

---

Phase 3 — VERIFY LIVE

Mark todo[2] in_progress.

3a. Verify live site.

REQUIRED SUB-SKILL: rune:browser-pilot
→ Invoke `browser-pilot` with the deployed URL.
→ browser-pilot checks: page loads (HTTP 200), no console errors, critical UI elements visible.
→ Capture: screenshot, status code, load time, any JS errors.

Error recovery:
  If browser-pilot returns non-200 or JS errors:
    → Report: "LIVE VERIFY FAILED: [details]"
    → Do NOT proceed to Phase 4
    → Present screenshot + error log to user

3a.5. Apply post-deploy health thresholds.

Use the metric-based rollback decision matrix defined in rune:deploy Step 4.5. Compare error rate, latency (p95), and availability against pre-deploy baseline for 15 minutes:

• ALL metrics ADVANCE → proceed to Phase 4
• ANY metric HOLD → extend monitoring to 30 minutes, alert user before proceeding
• ANY metric ROLLBACK → stop pipeline, invoke rune:incident, do NOT proceed to marketing

This step bridges deploy's numeric thresholds into the launch pipeline — launch must not proceed to marketing while the deployment is unhealthy.

3b. Setup monitoring.

REQUIRED SUB-SKILL: rune:watchdog
→ Invoke `watchdog` with: url=[deployed URL], interval=5min, alert_on=[5xx, timeout].
→ watchdog configures health check endpoint monitoring.
→ Capture: monitoring confirmation + health endpoint path.

Mark todo[2] completed when live verification passes and monitoring is active.

---

Phase 4 — MARKET

Mark todo[3] in_progress.

4a. Generate marketing assets.

REQUIRED SUB-SKILL: rune:marketing
→ Invoke `marketing` with: project context, deployed URL, key features.
→ marketing generates:
    - Landing page hero copy (headline, subheadline, CTA)
    - Twitter/X announcement thread (3-5 tweets)
    - LinkedIn post
    - Product Hunt tagline + description
    - SEO meta tags (title, description, og:image alt)
→ Capture: all generated copy as structured output.

4b. Optional — launch video.

If user requested video content:
  REQUIRED SUB-SKILL: rune:video-creator
  → Invoke `video-creator` with: deployed URL, feature list, target platform.
  → Capture: video script + asset manifest.

Mark todo[3] completed when all requested assets are generated.

---

Phase 5 — ANNOUNCE

Mark todo[4] in_progress.

Present all assets to user in structured format. Do not auto-publish — user approves before posting.

Present:
  - Deployed URL (clickable)
  - Monitoring status
  - All marketing copy blocks (ready to copy-paste)
  - Video script (if generated)
  - Next steps checklist

Mark todo[4] completed.

---

Constraints

1. MUST pass ALL tests before any deploy attempt — zero exceptions
2. MUST pass sentinel security scan before deploy — no CRITICAL findings allowed
3. MUST have rollback plan documented before deploying to production
4. MUST NOT deploy and run marketing simultaneously — deploy first, verify, then market
5. MUST verify deploy is live and healthy before triggering marketing skills

Mesh Gates

Gate	Requires	If Missing
Test Gate	verification output showing all green	Run rune:verification first
Security Gate	sentinel output with no CRITICAL findings	Run rune:sentinel first
Deploy Gate	Successful deploy confirmation before marketing	Deploy first

Output Format

## Launch Report
- **Status**: live | failed | partial
- **URL**: [deployed URL]
- **Tests**: [passed]/[total]

### Deployment
- Platform: [Vercel | Netlify | custom]
- Build: [success | failed]
- URL: [live URL]

### Monitoring
- Health endpoint: [path]
- Check interval: 5min
- Watchdog: active | failed

### Marketing Assets
- Hero copy: [ready | skipped]
- Twitter thread: [ready | skipped]
- LinkedIn post: [ready | skipped]
- Product Hunt: [ready | skipped]
- SEO meta: [ready | skipped]
- Launch video: [ready | skipped]

Returns

Artifact	Format	Location
Deploy status + live URL	Inline (Launch Report)	Emitted at session end
Marketing assets (copy, social, SEO)	Markdown (inline)	Generated by rune:marketing, presented in Phase 5
Release checklist	Markdown (inline)	Shown in Announce phase
Monitoring confirmation	Inline	Watchdog setup output
Launch Report	Markdown (inline)	Emitted at end of session

Sharp Edges

Known failure modes for this skill. Check these before declaring done.

Failure Mode	Severity	Mitigation
Attempting deploy with failing tests or TypeScript errors	CRITICAL	HARD-GATE blocks this — pre-flight must be 100% green
Running marketing before deploy verified live	HIGH	Constraint 4: deploy → verify HTTP 200 → THEN market. Never simultaneous
No rollback plan before production deploy	MEDIUM	Constraint 3: document rollback strategy before running deploy command
Platform auto-detected incorrectly (wrong deploy command)	MEDIUM	Verify platform config files before running — ask if ambiguous
Marketing assets generated from assumptions rather than scout output	MEDIUM	Step 1 requires scout to run — copy based on actual features, not assumptions

Done When

• Pre-flight PASS: all tests, types, lint, build, and sentinel green
• Deploy command succeeded with live URL captured
• Live site returns HTTP 200 (curl or browser-pilot confirmed)
• watchdog monitoring active on deployed URL
• All requested marketing assets generated (or skipped with reason)
• User presented with all assets before any publishing
• Launch Report emitted with URL, monitoring status, and asset list

Cost Profile

~$0.08-0.15 per launch. Sonnet for coordination, delegates to haiku for scanning.

Scope guardrail: Do not publish marketing assets or trigger external announcements unless explicitly delegated by the parent agent.