RENAR experimental

Core approach

Treats business, system, and task requirements as the source of expected behavior. Specifications and positive/negative test cases form an approved description set, with verification tied to both its version and a product version. Human approvals govern artifact transitions; the storage backend must provide the standard's history, review, version-pinning, and attribution capabilities.

Workflow

1. Register the client's statement of work (TZ) and review it for gaps and contradictions
2. Where clarification is needed, prepare an internal adaptation (ADAPT) and a client/vendor clarification protocol (ACTZ)
3. Derive business requirements (BR), system requirements (SR), specifications (SPEC), and task requirements (TR)
4. Create positive and negative test-case norms (TC) and approve the complete description-set version at QG-0
5. Implement and admit test cases at QG-1; use a different agent to implement the product
6. Run the version's admitted test cases against the product and record QG-2 verification
7. Derive isolated acceptance tests from the effective statement of work and obtain the applicable human approvals

Supported tools

• Any AI coding agent

Notes

RENAR (Requirements Engineering & Normative Adaptive Regulation), by Vadim Soglaev and Andrey Yumashev, was included alongside SENAR in submission #83. It merits its own framework entry: its documentation explicitly defines a standalone SDD methodology and includes an English quickstart, artifact templates, and an operational edition for agents, RENAR-AGENT-EN.md.

The reviewed v1.1 documentation distinguishes the architect-approved internal ADAPT from the ACTZ signed by the client and vendor. The agent guide and normative lifecycle chapter provide the detail behind the shorter summary. QG-1 here admits a test-case implementation to runs; it is not simply a general code-implementation phase.

SENAR provides the complementary supervision standard. RENAR’s experimental rating reflects the limited independently verifiable adoption evidence reviewed. Detailed normative requirements are not proof that a particular agent or backend enforces them.

Sources

• RENAR overview and release history
• English quickstart and artifact templates
• Working with an AI agent and human approval boundaries
• Normative lifecycle and quality gates

Added Sep 30, 2026 · Assessment last reviewed Sep 30, 2026 · How we track