Interested in working together?
Side Project
Shipwise — Ship Your Webapp Wisely
Claude CodePluginAI ToolingOpen SourceSecurityDevOpsLaunch Lifecycle
VIEW PROJECT →Problem
AI coding assistants write solid code but don't know about your project's launch readiness. They won't warn you about missing error tracking before a deploy, suggest CSRF protection when you write auth code, or remind you that you need a privacy policy. Every launch, developers rediscover the same gaps — security headers, health endpoints, legal pages, billing webhooks — because nothing tracks readiness across sessions.
Ownership
Sole creator. Designed the plugin architecture, 15 domain skills, 4 lifecycle hooks, 60+ reference documents, QA test framework, and landing page. Conducted competitive analysis against 25+ ecosystem repos. Open-sourced on GitHub under MIT license.
Architecture
15 domain skills organized into 4 lifecycle phases (Design → Build → Ship → Grow), each with a SKILL.md file and references/ directory containing copy-paste templates. A JSON state file (.claude/shipwise-state.json) tracks launch readiness across sessions. Four bash hooks fire at lifecycle moments: session start (injects readiness score), post-edit (contextual whispers for auth/billing/infra code), pre-deploy (P0 gap gate), and session end (auto-progress detection). An orchestrator skill (launch-assess) runs the initial diagnostic interview, scans the codebase, and generates the personalized checklist. Experience-level calibration adapts all output for beginner, intermediate, or senior developers.
Highlights
- •15 domain skills + 1 orchestrator covering idea validation through post-launch growth ops
- •45+ reference documents — every one has at least one copy-paste template
- •Managed auth provider detection: if you use Clerk or Auth0, the security skill skips irrelevant items and focuses on what you're responsible for
- •SaaS boilerplate comparison: maps 5 open-source starters to checklist items they already satisfy
- •Multi-tenancy patterns: shared DB, Postgres RLS, schema-per-tenant, Clerk Organizations — all with copy-paste code
- •Before/After validation: same auth prompt scored 18/40 without Shipwise, 36/40 with it — 24 specific security improvements
- •12-scenario QA suite with 3 test fixtures (beginner, mid-level, senior archetypes), all 12 passing
- •Deploy gate warns before deploying with P0 gaps incomplete
- •Codebase auditor detected 9/9 tech components in realistic SaaS fixture
Deployment
Distributed as a Claude Code plugin — installable via terminal. All state stored locally in .claude/ per project. GitHub Pages landing page. GitHub Actions CI for plugin structure validation. Multi-agent compatible via convert.sh (Codex, Cursor, Gemini CLI, Windsurf, Aider, OpenCode, Augment, Antigravity). No backend, no auth, no infrastructure.
Impact
Only lifecycle-aware plugin in the Claude Code ecosystem. Covers ground no individual skill does: the intersection of engineering, security, legal, business formation, and growth — orchestrated with phase-aware routing and persistent readiness tracking. Validated with the strongest proof point in the agent skills space: a reproducible Before/After test showing 2× security improvement on the same prompt.