Agent skill · software engineering · jeremylongshore
shipwright-pipeline
Autonomous app builder that converts plain-English descriptions into
Why this skill is useful
Adds executable scripts for building and deploying full-stack applications from plain-English descriptions, which the AI cannot generate on its own.
What it needs
Requires node, product-agent, python installed locally. About 2k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Shipwright Pipeline Overview Shipwright converts a plain-English app description into a fully built, tested, and deployment-ready application. It delegates execution to product-agent, an autonomous 9-phase build engine available on PyPI. Prerequisites Python 3.10+ available in PATH. product-agent installed (pip install product-agent). Node.js 18+ for JavaScript/TypeScript stacks. Supported Stacks Next.js + Supabase — Full-stack with auth, database, and edge functions Next.js + Prisma — Full-stack with type-safe ORM SvelteKit — Lightweight full-stack with Svelte Astro — Content-focused static and hybrid sites Instructions 1. Gather the app description from the user. Ask clarifying questions if the description is vague. 2. Confirm the target stack. If none specified, recommend based on the app type: Data-heavy with auth: Next.js + Supabase API-first with complex models: Next.js + Prisma Lightweight interactive: SvelteKit Content or marketing site: Astro 3. Run product-agent with the app description and selected stack. 4. Monitor the 9-phase pipeline: Intake, Architecture, Scaffold, Implement, Test, Integrate, Polish, Validate, Ship. 5. Report results to the user including test summary and any warnings. Output A complete, buildable project directory with all source code, tests, and configuration. A test report summarizing pass/fail counts. Build verification output confirming the project compiles and starts. Error Handling If product-agent is not installed, prompt the user to install it with pip install product-agent. If a phase fails, report the phase name, error message, and suggested fix. If the selected stack is not supported, list available stacks and ask the user to choose. Examples Build a SaaS dashboard: Add features to an existing project: Scaffold a content site: Resources Shipwright on GitHub product-agent on PyPI See ${CLAUDESKILLDIR}/references/examples.md for usage examples.
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/shipwright-pipeline--23d314