Chimera: An Autonomic, Intent-Centric Architecture for Verifiable Decentralized Intelligence

Abstract

Current decentralized architectures, while powerful, operate in silos. Intent-centric protocols solve for user expression but create trust assumptions for off-chain solvers. 1 Verifiable computation markets address this trust but lack a native framework for complex, goal-oriented coordination. 2 Formal verification methods provide security for individual components but struggle with the emergent complexity of their composition. 3 This paper introduces Chimera, a novel protocol architecture that unifies these disparate paradigms. Chimera integrates generalized intent-centricity, a decentralized market for verifiable AI agents (zk-Agents), and autonomic governance inspired by closed-loop control theory, all within a framework of provable composability. We argue that by treating all network actions-from swaps to governance-as intents fulfilled by verifiable AI agents, and by governing the system with an autonomic management layer that monitors and adapts its own parameters, Chimera enables a new class of self-organizing, intelligent, and provably secure decentralized applications. We present the full architectural design and demonstrate its power through case studies in verifiable AI-driven finance and autonomic public goods funding.