Independent · June 2025 — present
ROME
Requirements-Oriented Multi-agent Engineering
A methodology framework that orchestrates specialised Claude Code agents across a complete SDLC pipeline — from structured requirements capture through to working code, with full traceability at every step.
By the numbers
10+
Coordinated agents
13
AORDL fields / requirement
40+
Reusable skills
100%
Forward & backward traceability
How it works
Four stages, one traceable pipeline.
Each stage hands off to the next through a structured artefact, not a conversation — so every requirement can be traced forward to the code that satisfies it, and every line of code traced back to the requirement that demanded it.
AORDL capture: 13 mandatory fields per requirement, no ambiguity carried forward.
Domain modelling and structured business analysis, AI-assisted, human-reviewed.
Design elaboration from the analysed requirements, with the same traceability held throughout.
Branch-per-phase delivery, fan-out/fan-in agent execution, PR-based quality gates.
What's inside
Multi-agent pipeline design using subagents with fan-out/fan-in parallel execution. Plugin architecture bundles skills, hooks, MCP servers and subagents into distributable packages.
Branch-per-phase Git strategy with PR-based quality gates, so every phase of the pipeline lands as a reviewable, revertible unit.
AORDL — Actor-Oriented Requirements Definition Language — structures every requirement the same way, enabling forward traceability (requirement → feature → story → spec → code) and backward traceability the other way.