flow-alpha

Orchestrates end-to-end alpha ADR development with stage-gated verification and human checkpoints.

12|2|Updated Nov 27, 2024
One-click install
npx skills add https://github.com/Pionpill/retikz --skill flow-alpha
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: flow-alpha
Source: https://github.com/Pionpill/retikz/tree/main/.agents/skills/flow-alpha
Command: npx skills add https://github.com/Pionpill/retikz --skill flow-alpha

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

flow-alpha orchestrates the end-to-end, stage-gated alpha feature development workflow for a single ADR or multiple ADRs, preventing unreviewed or incomplete changes from progressing.

Core Features & Use Cases

  • Two execution modes with a blocking human choice: single-ADR (run 5 stages then stop) or batch worktree (N ADRs, each with its own worktree + branch, designed for offline runs).
  • Stage-gated quality controls across 5 hard phases: design → implementation (with Spec-First for red changes) → adversarial self-test → bilingual documentation + demo/API coverage → wrapup with changelog/roadmap/ADR acceptance.
  • Dependency-aware parallelism for batch mode: analyzes ADR file-scope and explicit dependencies to decide parallel worktrees vs stacked chains, then enforces strict per-worktree isolation and review gating.

Quick Start

Ask the AI to run the alpha flow for ADR-XX in single mode after you confirm whether this run is single-ADR or batch worktrees.

Frequently Asked Questions about flow-alpha

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

FAQPage Schema
How do I enforce quality gates across an ADR workflow end-to-end?▼

You can enforce quality gates across an ADR workflow by orchestrating stage-gated development spanning design, implementation, adversarial testing, documentation, and wrapup phases with strict human review checkpoints.

What's the best way to manage multiple ADRs in parallel git worktrees?▼

Batch worktree mode manages multiple ADRs by analyzing file-scope and explicit dependencies to decide parallel worktrees versus stacked chains, enforcing strict per-worktree isolation and review gating for each branch.

How does spec-first TDD work when implementing ADR changes?▼

Spec-first TDD for ADR implementation requires writing failing tests for red changes before writing code, ensuring every implementation phase begins with verifiable specifications and stage-specific verification like lint, test, and tsc checks.

Can I run a single ADR through all development stages without batch processing?▼

Single-ADR execution mode runs five sequential stages—design, implementation, adversarial self-test, bilingual documentation, and wrapup—then stops, requiring a blocking human selection to confirm this mode before execution begins.

Why does adversarial testing matter in alpha-stage feature development?▼

Adversarial testing matters in alpha-stage development because it acts as a dedicated quality gate to challenge implementation robustness through self-testing before progressing to documentation and final ADR acceptance.

What are the safety limitations when running ADR workflows in git worktrees?▼

Safety limitations include strict non-merge and non-push rules, blocking human selection of execution mode, and per-worktree isolation, preventing unreviewed or incomplete changes from progressing across any stage gate.