yolo-ship

Coordinate end-to-end delivery from planning to a CI-green pull request.

1|Updated Apr 23, 2026
One-click install
npx skills add https://github.com/project-ax/ax-next --skill yolo-ship
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: yolo-ship
Source: https://github.com/project-ax/ax-next/tree/main/.claude/skills/yolo-ship
Command: npx skills add https://github.com/project-ax/ax-next --skill yolo-ship

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It eliminates slow, back-and-forth delivery by taking a feature request from plan to reviewed code and a CI-green pull request automatically, with clear ownership expectations and safety guardrails.

Core Features & Use Cases

  • Autonomous end-to-end delivery loop: brainstorm, design, implement (via subagents), run full pre-PR gates, and drive a clean review before opening the PR.
  • Prevents “almost done” outcomes: enforces pnpm build, pnpm test, and lint as a hard gate, then ensures the branch is reviewed locally before the PR exists.
  • Traceable decisions + tracked follow-ups: logs non-trivial decisions to .claude/memory/decisions.md and returns deferred work as TO DO items rather than silently dropping it.
  • Isolation-first workflow: uses dedicated worktrees (never the main checkout) to avoid clobbering other agents’ work.
  • Use case: when you need “implement X end to end” or “ship this task” with minimal supervision, this skill orchestrates the full implementation and delivery process through merge (standalone) or handoff (orchestrated).

Quick Start

Use yolo-ship when you want to implement a feature end to end and open a review-clean pull request with CI passing, asking for as little supervision as possible.

Frequently Asked Questions about yolo-ship

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I implement a feature end to end and open a pull request with minimal supervision?

To implement a feature end to end and open a pull request autonomously, use a delivery loop that handles planning, subagent implementation, local branch review, and CI verification before opening the PR.

Can I run autonomous code delivery without affecting other agents' work in the main checkout?

Autonomous code delivery can run safely without affecting other agents by enforcing worktree isolation, which uses dedicated worktrees instead of the main checkout to prevent clobbering parallel work.

How do I prevent "almost done" pull requests from failing CI checks?

To prevent pull requests from failing CI checks, enforce pnpm build, pnpm test, and lint as hard gates, then run a whole-branch local review before the PR exists to ensure CI-green delivery.

What is the best way to track deferred work and non-trivial decisions during autonomous code delivery?

To track deferred work and non-trivial decisions during autonomous code delivery, log decisions to a memory file and return deferred work as TO DO items rather than silently dropping them.

Does autonomous PR delivery work with pnpm verification and local code review?

Autonomous PR delivery works with pnpm verification and local code review by running pnpm build, test, and lint as hard gates, then orchestrating a clean local review before opening the pull request.

When should I not use an autonomous delivery loop for shipping code changes?

You should not use an autonomous delivery loop when your task requires board routing manipulation, lacks clear ownership expectations, or needs manual supervision outside the isolated worktree workflow.