SAMF- Deterministic Moscow guardrails for LLM multi-agent loops

A new open-source project called SAMF introduces machine-readable validation contracts for LLMs, translating MoSCoW prioritization into deterministic guardrails for multi-agent loops. The tool aims to eliminate brittle multi-hop reasoning, prevent loop breaches, and optimize token costs by enforcing explicit constraints.

Structural MoSCoW contracts for deterministic LLM validation. SAMF provides machine-readable validation contracts for LLMs, translating project management MoSCoW prioritization into strict alignment guards. Slashes Prompt Instability: Eliminates brittle multi-hop reasoning. Prevents Loop Breaches: Keeps multi-agent reasoning loops from breaking. Optimizes Token Costs: Reduces RAG token waste by enforcing explicit constraints. python from samf import SAMFContract, samf contract Define your strict execution contract guard contract = SAMFContract must have= "JSON", "status", "user id" , should have= "timestamp" , wont have= "error", "failed" Enforce it seamlessly with a decorator @samf contract contract=guard contract def call billing agent : Your LLM generation / LangGraph node logic here return '{"status": "success", "user id": 4002, "format": "JSON"}' response = call billing agent print "Output validated successfully " --- Disclaimer: This is an independent open-source hobby project developed entirely on personal time and hardware. It is not affiliated with, sponsored by, or endorsed by my employer.