super-build

Dispatch GitHub Project issue cards in isolated git worktrees and merge results.

Updated May 20, 2026
One-click install
npx skills add https://github.com/samuelcastro/skills --skill super-build
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: super-build
Source: https://github.com/samuelcastro/skills/tree/main/super-build
Command: npx skills add https://github.com/samuelcastro/skills --skill super-build

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires git, gh, jq, claude -p, and includes references (resource) and scripts (resource) components.

What problem does it solve?

It eliminates the slow, manual process of draining a GitHub Project queue and running issue-by-issue “builder” work across many cards, while keeping merges and issue state updates consistent.

Core Features & Use Cases

  • Parallel headless worker dispatch: Executes each curated GitHub Project Ready (or Bug in QA mode) issue as an isolated worker in its own git worktree, with a max-parallel throttle to keep runs efficient and controllable.
  • Safe merge-and-close workflow: Merges worker branches back to the base branch, closes the issue on success, and advances the card status to Done.
  • Failure handling and human gates: Halts on worker failures or merge conflicts, supports intentional partial progress via WIP-PARTIAL, and requires human handling for human-gated cases.
  • Issue contract enforcement: Uses a worker preamble to require correct close/WIP commit formats and to respect acceptance-criterion checkboxes without trying to “game” completion.

Quick Start

Ask your agent to run the build loop by saying: run super-build to execute all GitHub Project cards in the Ready column.

Frequently Asked Questions about super-build

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

FAQPage Schema
How do I execute GitHub Project issues in parallel using isolated git worktrees?

This Skill drains GitHub Project backlogs by dispatching each Ready issue as an isolated worker in its own git worktree, enforcing a max-parallel throttle for controllable concurrency. It then reconciles branches back to the base.

What is the best way to automate merging and closing GitHub issues after autonomous builds?

Automating merge-and-close workflows is handled by merging worker branches back to the base branch, closing the GitHub issue on success, and advancing the project card status to Done. It halts on failures or merge conflicts.

Do I need a headless coding-agent CLI to run autonomous build queues?

Yes, running autonomous build queues requires a headless coding-agent CLI like 'claude -p', alongside git, gh, and jq. These dependencies power the dispatcher/exit-code contract for worker execution and status reconciliation.

How does parallel issue execution handle worker failures and partial progress?

Parallel issue execution handles failures by halting on worker errors or merge conflicts, while supporting intentional partial progress via the WIP-PARTIAL commit format. It also enforces human-gated safety labels for cases requiring manual review.