Structure is not a guideline; it is an agreement. The value of a language lies not in the freedom it provides, but in the ambiguity it removes.
Requirements as Code for Systems Engineering
RQMT is an open-source framework for writing, validating, and managing requirements using a structured language designed for both humans and machines.
The project is built around the idea that requirements should be treated similarly to source code:
- Written in a formal syntax
- Parsed into an Abstract Syntax Tree (AST)
- Validated automatically
- Version controlled with Git
- Supported by IDE tooling
- Traceable across development artifacts
RQMT aims to bridge the gap between stakeholders, systems engineers, and developers through a common, machine-processable representation of requirements.
Why RQMT?
Requirements are often written as free-form documents.
While this approach is flexible, it introduces several challenges:
- Ambiguity
- Inconsistent structure
- Difficult traceability
- Limited automation
- Poor integration with development workflows
Modern software development benefits from compilers, static analysis, IDEs, and automated tooling.
RQMT applies similar principles to requirements engineering.
Development Setup
Prerequisites
- Node.js
- CMake
- A C++ compiler
- VS Code (recommended)
Installation
npm install
npm run compile
Build
Use the VS Code CMake extension or:
cmake -S server/src -B build
cmake --build build
Example
req-1 User Authentication:
[to_implement]
The system shall authenticate users before granting access.
fr{login}-2 Login:
The system shall provide a login interface.
@draft
nfr-3 Security:
Authentication credentials shall be stored securely.
The document can be parsed into a structured representation:
Document
└─ req-1 User Authentication
├─ fr-2 Login
└─ nfr-3 Security
Core Concepts
Structured Requirements
Requirements are written using a simple hierarchical syntax.
req-10 Navigation:
The robot shall navigate autonomously.
fr-11 Localization:
The robot shall estimate its position.
Abstract Syntax Tree (AST)
RQMT documents are parsed into an AST that can be analyzed and processed automatically.
Requirements as Code
Requirements become machine-processable artifacts that can participate in development workflows.
Traceability
RQMT is designed with traceability in mind.
Requirements can be uniquely identified, referenced, validated, and linked to:
- Requirements
- Design artifacts
- Source code
- Test cases
- Documentation
Architecture
RQMT Document
│
▼
Lexer
│
▼
Parser
│
▼
AST
│
├── Diagnostics
├── Validation
├── Traceability
└── Tooling
Motivation
RQMT was originally motivated by the challenges of systems engineering in multidisciplinary domains such as robotics, where requirements often serve as the primary interface between:
- Stakeholders
- Systems engineers
- Software engineers
- Mechanical engineers
- Electrical engineers
- Verification and validation teams
RQMT provides a common, easy to learn, structured representation that can be understood by humans and processed by tools.
Status
RQMT is currently in active development, in an early stage. The language specification and tooling are evolving and may change significantly before the first stable release.
Proof of Concept
Version: 0.1.0.x
A rudimentary implementation of the LSP can be run as VSC extension:
Implemented:
- Lexer
- Parser
- AST generation
- Incremental parsing
- Syntax diagnostics
- Semantic tokens and syntax high-lighting
- Go-to-definition
reference→alias
In progress:
- Semantic diagnostics
- Auto-completion
- Automatic ID generation
Roadmap
v0.2: Core Semantics
- Reference resolution
- Semantic diagnostics
- Alias validation
- Tag syntax
- Decorators syntax
- Link tests and implementations
v0.3: Modularity and Usability
- Cross-file references
- Workspace indexing
- Auto-ID workflow
- Hover information
- Formatting
v0.4: Advanced Tooling
- Workspace indexing (optimized)
- Rename symbol
- Traceability views
- Export to Markdown/HTML
- Configuration files (.rqmtconfig.json)
v0.5: Automation and CI/CD
- CI/CD Integration
- Static Analysis (CLI: rqmt lint)
- Flags: blocking releases, draft
- Basic AI-assisted analysis (ambiguity, completeness)
v0.6: AI for Content
- AI-assisted analysis (conflicts, standards compliance)
- Collaborative review tools (GitHub/GitLab integration)
v0.7: Advanced AI and Collaboration
- AI for domain-specific checks (e.g., ISO 26262 compliance)
- Plugin system (custom rules/checks)
v1.0: Stability and Production
- Stable language specification
- Complete LSP support
- Production-ready tooling
Documentation
Contributing
Contributions, feedback, discussions, and feature requests are welcome.
License
MIT License