rlm-workflow

What problem does it solve?

RLM-workflow enforces a disciplined, reproducible repository-driven development process so that requirements, plans, tests, and QA live in durable files rather than transient chat context. It prevents context rot, accidental main-branch pollution, and undocumented shortcuts by requiring sequential phase gates, immutable locks, and explicit user sign-off for manual QA.

Core Features & Use Cases

• Orchestrates end-to-end RLM phases (worktree isolation, AS-IS analysis, optional root-cause debugging, TO-BE planning, TDD implementation, review, parallel testing, manual QA, and global updates).
• Enforces TDD discipline, Coverage and Approval gates, LockHash (SHA-256) verification, addenda policy, and TODO completion rules to produce auditable run artifacts.
• Supports parallel subagent execution with automatic fallback to sequential mode, plus utilities to initialize runs, lint artifacts, verify locks, and report run status.
• Use case example: implement a feature or fix a bug as a resumable, single-command RLM run where each phase is documented, validated, and locked for auditability.

Quick Start

Implement requirement 2026-02-24-add-oauth to execute the full RLM run for that requirement.