AI Requirements Quality Engineering
Find the faults in your requirements before they become bugs.
REQQA reads every requirement and user story, runs eleven analysis dimensions, and proposes the changes that close the issues — so the spec is ready before code is written.
Reset password — channel?
Users must be able to reset their password through email or SMS at any time.
Analyse
11-step requirement analysis · 7-step DeFOSPAM for stories.
Generate
Mission → requirements → features → Gherkin scenarios.
Synthesise
Concrete edits, placeholders and advisories from issues.
Converge
Track quality across cycles. Approve when issues clear.
The framework
Eleven dimensions.
One reviewable list of issues.
Where most teams rely on a reviewer spotting problems under time pressure, REQQA applies a repeatable analytical framework to every requirement — and does it in minutes.
Read the full paper →Definitions
Goals & Context
Features
Interfaces
Rules
Entities & Data
Conditions
Boundaries
Quality Attributes
Ambiguity
Missing Elements
7 dimensions for Gherkin user stories.
The convergence loop
Write. Analyse. Improve.
Re-analyse. Ship.
REQQA tracks how a specification improves across analyse-improve cycles. Convergence history shows whether your spec is settling toward quality — or stalling.
- 01M-VMission
- 02R-*Requirements
- 03R-FFeatures
- 04D-*Stories
- 05ISSIssues
- 06SYNSynthesis
- 07CNVConverge
Convergence · APP 28
Issues across analysis cycles
22 → 1
six cycles · one shippable spec
Who it's for
Built for the people who carry specification quality.
Requirements Authors
Structured environment for writing, versioning, improving — with AI feedback on completeness and quality.
QA Leads & Test Architects
Multi-dimensional analysis to surface ambiguities and missing scenarios before development begins.
Team Leads & Managers
Visibility into specification readiness, convergence metrics and portfolio-wide issue tracking.
AI Builders
Token-authenticated public API for fetching scoped specs, submitting plans, posting feedback.
Development Teams
Higher-quality handoffs with fewer ambiguities, complete glossaries and validated acceptance criteria.
Evaluators & Trainers
Demo catalogue of pre-built applications to walk the full workflow without manual authoring.
Dark factory · Public API
Hand off to an AI builder — without losing the audit trail.
A token-authenticated REST API plus a coordinated set of Claude Code skills. Builders fetch a scoped release, negotiate a plan, post mid-build feedback, and submit an as-built record. The analyst stays in control of approval; the build cycle is autonomous in between.
See the build loop →{
"plan_version": 3,
"approach": "...",
"work_breakdown": [ ... ],
"assumptions": [ ... ],
"blockers": [],
"queries": [],
"unmitigated_risks": []
}What makes REQQA different
Analytical, not generative-first
Most AI tools generate content. REQQA's core value is finding what's wrong, missing or ambiguous.
A proven methodology
DeFOSPAM and the 11-dimension framework — structured analytical techniques, not a prompt.
Closes the loop
Issues → synthesis → concrete reviewable changes → re-analysis → handover → build → acceptance.
AI builders are first-class
Public API + skill set make REQQA a platform AI agents can read, write to, and coordinate against.
Builds institutional knowledge
Every run enriches the glossary, identifies features, proposes tags, builds traceability.
Measures progress
Convergence shows whether specs are getting better. Release reviews show whether the build cycle is too.
Ready when you are
Specifications that ship.
Eleven AI analysis steps. One reviewable list of issues. Ready before code.