Alignment Contracts for Agentic Security Systems

TL;DR

This paper introduces alignment contracts as a formal framework to specify and enforce behavioral constraints on agentic security systems, ensuring controlled effects within defined boundaries.

Contribution

It formalizes alignment contracts with semantics, satisfaction criteria, and composition rules, enabling modular and decidable enforcement of behavioral constraints in security workflows.

Findings

Framework supports web-focused agentic security workflows.

Formal semantics and safety properties are established for effect traces.

Enforcement guarantees are provided under an explicit effect observability assumption.

Abstract

Agentic security systems increasingly combine LLM planners with tools that can discover, validate, and report vulnerabilities. This creates an asymmetric control problem: the system should retain strong offensive capability inside an authorized engagement, while the same capabilities must be denied outside scope. Existing guardrails provide useful policy controls, but they do not make this boundary a first-class formal contract over observable effects. We introduce alignment contracts, a framework for specifying and enforcing behavioral constraints over observable effect traces. A contract defines scope, allowed and forbidden effects, resource budgets, and disclosure policies. We give the language finite-trace semantics, characterize satisfaction as a safety property with finite violation witnesses, develop refinement and one-way composition rules for modular contract engineering, and show that admissibility checking is decidable. We instantiate the framework for web-focused agentic security workflows and show how the same structure extends to other effect profiles. Under an explicit Effect Observability Assumption, where all $\SigmaEff$-effects are mediated, the soundness theorem quantifies over the agent model and gives guarantees for mediated $\SigmaEff$-effects, including enforcement soundness for monitor-realized traces. We also state an assumption-lifted adaptation result and formalize limits through undecidability transfer and observability-boundary theorems. A Lean 4 artifact checks the formal core theorems used by the paper.