SENAR experimental

Core approach

SENAR Core treats a task's goal, acceptance criteria, negative scenarios, and scope as the specification. A Start Gate checks those inputs; a Done Gate checks the current work against recorded evidence, requirement-derived tests, a risk-based review checklist, and captured knowledge. The wider standard adds team roles, ceremonies, and governance.

Workflow

1. Define the goal, independently verifiable acceptance criteria, and at least one negative scenario
2. Set scope boundaries and pass the Start Gate (QG-0) before writing code
3. Implement with an AI coding agent and verify each criterion against recorded evidence
4. Derive tests from requirements and review the work using the appropriate checklist tier
5. Record decisions, dead ends, and known issues; pass the Done Gate (QG-2) against the current version
6. Track first-pass success and dead-end rate against locally measured baselines

Supported tools

• Any AI coding agent
• Claude Code
• Cursor
• GitHub Copilot
• Windsurf

Notes

SENAR (Supervised Engineering & Normative AI Regulation), by Andrey Yumashev and Vadim Soglaev, qualifies as a documented SDD methodology: tasks define expected behavior before implementation, and completion depends on evidence against that definition. This assessment uses the repository’s v1.5 Core rather than treating the website’s older v1.4 overview as the current standard.

TAUSIK is its reference implementation and declares SENAR Core conformance. RENAR is a complementary, independently adoptable requirements methodology. They have separate entries because the standard, requirements lifecycle, and implementing tool serve different purposes.

The experimental rating reflects the limited independently verifiable adoption and outcome evidence reviewed, rather than the breadth of the documentation. The quickstart itself calls for independent replication of its reported results.

Sources

• SENAR repository overview
• SENAR Core v1.5: rules, gates, and metrics
• Quickstart and single-project evidence caveat
• Tool integration guide

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