skills.rest
spideynolove
spideynoloveCOMMUNITY
Explore all skills by this author →
Website𝕏@spideynoloveGitHub

rlm-workflow

Orchestrates phase-gated software development runs with locked artifacts, TDD discipline, and git worktree isolation.

2|3|Updated Mar 13, 2026
Search Tags:#tdd#root cause analysis#ai development#workflow orchestration#git worktree#phase gates#artifact locking
One-click install
npx skills add https://github.com/spideynolove/claude-dotfiles --skill rlm-workflow-spideynolove
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: rlm-workflow
Source: https://github.com/spideynolove/claude-dotfiles/tree/main/.agents-global/skills/rlm-workflow
Command: npx skills add https://github.com/spideynolove/claude-dotfiles --skill rlm-workflow-spideynolove

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? It prevents context rot and requirement drift in AI-assisted development by storing requirements, plans, and analysis in versioned repo files instead of chat context, enforcing strict phase gates so no step is skipped. ## Core Features & Use Cases - Stage-Gated RLM Workflow: Runs Phases 0-7 (worktree setup, requirements, AS-IS analysis, root cause debugging, planning, TDD implementation, code review, testing, manual QA, global state updates) with SHA-256 lock hashes and hard gates blocking progression until prior phases pass. - TDD and Debugging Discipline: Enforces RED-GREEN-REFACTOR cycles with a mandatory TDD Compliance Log and an optional Phase 1.5 systematic root cause analysis for bug fixes. - Parallel or Sequential Execution: Automatically detects subagent capability to run implementation and test suites concurrently, with a rigorous sequential fallback mode. - Use Case: A developer types "Implement requirement 'add-oauth'" and the workflow auto-resumes through analysis, planning, TDD implementation, and testing in an isolated git worktree, pausing only for manual QA sign-off before updating DECISIONS.md and STATE.md. ## Quick Start Ask the agent to implement requirement '<run-id>' after creating a requirements file at .codex/rlm/<run-id>/00-requirements.md.

Frequently Asked Questions about rlm-workflow

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

FAQPage Schema
How do I start an RLM workflow run?▼

Create a run folder at .codex/rlm/<run-id>/ containing a 00-requirements.md file, then invoke 'Implement requirement <run-id>'. The workflow auto-resumes through all phases, pausing only for manual QA sign-off in Phase 5.

How does RLM workflow enforce TDD discipline?▼

Phase 3 follows the Iron Law: no production code without a failing test first. Every implementation must document RED-GREEN-REFACTOR cycles in a TDD Compliance Log with timestamps and actual test output before the phase can be locked.

Does rlm-workflow work with Claude Code and Codex?▼

Yes, it runs on Claude Code, Codex, OpenCode, and Cursor. It automatically detects subagent capability (like Claude Code's Task tool) to enable parallel execution, falling back to sequential mode when subagents are unavailable.

Why does the workflow require git worktrees?▼

Phase 0 creates an isolated worktree at .worktrees/<run-id> to prevent main branch pollution and verify a clean test baseline before changes. Working directly on main requires explicit documented consent.

What happens if a phase artifact fails verification?▼

The lock chain hard gate stops all progression. The workflow resumes the earliest failing phase, repairs it until Coverage and Approval gates pass with a valid LockHash, then continues sequentially.

When should I use Phase 1.5 root cause analysis?▼

Use Phase 1.5 for bug fixes, test failures, or unexpected behavior. It enforces systematic debugging (error analysis, reproduction, data flow tracing, hypothesis testing) and must be locked before any fix planning begins.

Related Skills

Explore More Software Engineering Skills→
justinmoonjustinmoon

agents-md

Bootstrap and improve AGENTS.md files with agent collaboration best practices.

Community
Intermediate
jimehjimeh

merge

Merges a topic branch into the base branch across Git worktrees with conflict resolution.

Community
Intermediate
emcconnellemcconnell

team-qa

Coordinate QA team execution to produce test plans, cases, bug reports, and sign-off reports.

Community
Advanced
Pear-CommercePear-Commerce

gsd-quick

Execute small development tasks with atomic commits and STATE.md tracking.

Official
Intermediate