skills.rest
azusagasaku
azusagasakuCOMMUNITY
Explore all skills by this author →
GitHub

ralphinho-rfc-pipeline

Decomposes large features into RFC-driven work units with DAG orchestration and merge queues.

Updated May 19, 2026
Search Tags:#multi-agent#orchestration#quality gates#RFC#dag#merge queue#work units
One-click install
npx skills add https://github.com/azusagasaku/--claude-config --skill ralphinho-rfc-pipeline-azusagasaku
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ralphinho-rfc-pipeline
Source: https://github.com/azusagasaku/--claude-config/tree/main/skills/ecc/ralphinho-rfc-pipeline
Command: npx skills add https://github.com/azusagasaku/--claude-config --skill ralphinho-rfc-pipeline-azusagasaku

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Large features often exceed what a single agent pass can implement reliably, leading to broken integrations and unverifiable changes. This Skill splits oversized features into independently verifiable work units coordinated through a DAG, quality gates, and a merge queue. ## Core Features & Use Cases - DAG Decomposition: Breaks an RFC into work units with explicit dependencies, scope, acceptance tests, risk levels, and rollback plans. - Quality Gates per Unit: Enforces a research, plan, implement, test, review, and merge-ready pipeline for every unit. - Merge Queue & Recovery: Rebases unit branches, re-runs integration tests after each merge, and evicts or regenerates stalled units with narrowed scope. - Use Case: When asked to build a multi-module authentication overhaul, use this Skill to decompose the RFC into Tier 1-3 units, assign them to agents, validate each against acceptance tests, and merge them through the integration queue. ## Quick Start Use the ralphinho-rfc-pipeline skill to decompose this feature RFC into dependency-ordered work units and orchestrate their implementation through the merge queue.

Frequently Asked Questions about ralphinho-rfc-pipeline

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

FAQPage Schema
How do I split a large feature into smaller work units for AI agents?▼

Decompose the feature RFC into units each defining an id, depends_on list, scope, acceptance tests, risk level, and rollback plan. Order units in a DAG so dependencies are verified before downstream units start.

What is a merge queue in multi-agent development?▼

A merge queue serializes integration of completed work units by rebasing each unit branch on the latest integration branch and re-running integration tests after every merge. Units with unresolved dependency failures are never merged.

How are complexity tiers assigned to work units?▼

Tier 1 covers isolated file edits with deterministic tests, Tier 2 covers multi-file behavior changes with moderate integration risk, and Tier 3 covers schema, auth, performance, or security changes requiring the strictest review.

What happens when a work unit stalls during implementation?▼

The stalled unit is evicted from the active queue, its findings are snapshotted, and a narrowed unit scope is regenerated. The unit is then retried with updated constraints rather than blocking the pipeline.

When should I not use RFC decomposition for a feature?▼

Avoid it for small changes that fit a single agent pass, since DAG orchestration and merge queue overhead add coordination cost. It is designed for features too large to implement and verify in one pass.

Related Skills

Explore More Software Engineering Skills→
codetail-aicodetail-ai

pm

Creates structured PRDs with user stories, acceptance criteria, and scope boundaries.

Official
Advanced
openshiftopenshift
71

rebase-clean

Rebase a feature branch onto origin/main with conflict resolution and validation.

Official
Advanced
meshtasticmeshtastic
1.8k

testing-ci

Automate CI workflows for code quality, testing, and building across platforms.

Official
Advanced
robinxinrobinxin

ralphinho-rfc-pipeline

Automate RFC-driven feature decomposition into DAG work units with quality gates.

Community
Advanced