AgentAssay: Token-Efficient Regression Testing for Non-Deterministic AI Agent Workflows

TL;DR

AgentAssay introduces a token-efficient, statistically grounded framework for regression testing non-deterministic AI agent workflows, significantly reducing costs while ensuring reliable detection of regressions across multiple models and scenarios.

Contribution

It is the first to provide a comprehensive, token-efficient regression testing framework with statistical guarantees for non-deterministic AI agents, including novel metrics and offline analysis methods.

Findings

Behavioral fingerprinting achieves 86% detection power.

SPRT reduces testing trials by 78%.

Full pipeline achieves 100% cost savings.

Abstract

Autonomous AI agents are deployed at unprecedented scale, yet no principled methodology exists for verifying that an agent has not regressed after changes to its prompts, tools, models, or orchestration logic. We present AgentAssay, the first token-efficient framework for regression testing non-deterministic AI agent workflows, achieving 78-100% cost reduction while maintaining rigorous statistical guarantees. Our contributions include: (1) stochastic three-valued verdicts (PASS/FAIL/INCONCLUSIVE) grounded in hypothesis testing; (2) five-dimensional agent coverage metrics; (3) agent-specific mutation testing operators; (4) metamorphic relations for agent workflows; (5) CI/CD deployment gates as statistical decision procedures; (6) behavioral fingerprinting that maps execution traces to compact vectors, enabling multivariate regression detection; (7) adaptive budget optimization calibrating trial counts to behavioral variance; and (8) trace-first offline analysis enabling zero-cost testing on production traces. Experiments across 5 models (GPT-5.2, Claude Sonnet 4.6, Mistral-Large-3, Llama-4-Maverick, Phi-4), 3 scenarios, and 7,605 trials demonstrate that behavioral fingerprinting achieves 86% detection power where binary testing has 0%, SPRT reduces trials by 78%, and the full pipeline achieves 100% cost savings through trace-first analysis. Implementation: 20,000+ lines of Python, 751 tests, 10 framework adapters.