HadAgent: Harness-Aware Decentralized Agentic AI Serving with Proof-of-Inference Blockchain Consensus
Authors: Landy Jimenez, Mariah Weatherspoon, Bingyu Shen, Yi Sheng, Jianming Liu, Boyang Li
Organizations: Department of Computer Science and Technology, Kean University, Union, New Jersey, U.S. · Department of Computer Science and Engineering, University of Notre Dame, Notre Dame, Indiana, U.S. · Bellini College of Artificial Intelligence, Cybersecurity and Computing, University of South Florida, Tampa, Florida, U.S. · DefinanceX, U.S.
Proof-of-Work (PoW) blockchain consensus consumes vast computational resources without producing useful output, while the rapid growth of large language model (LLM) agents has created unprecedented demand for GPU computation. We present HadAgent, a decentralized agentic AI serving system that replaces hash-based mining with Proof-of-Inference (PoI), a consensus mechanism in which nodes earn block-creation rights by executing deterministic LLM inference tasks. Because verification requires only re-executing a single forward pass under identical conditions, cross-node verification operates at consensus speed. HadAgent organizes validated records into a three-lane block body with dedicated DATA, MODEL, and PROOF channels, each protected by an independent Merkle root for fine-grained tamper detection. A two-tier node architecture classifies secondary nodes as trusted or non-trusted based on historical behavior: trusted nodes serve inference results in real time through optimistic execution, while non-trusted nodes must undergo full consensus verification. A harness layer monitors node behavior through heartbeat probes, anomaly detection via deterministic recomputation, and automated trust management, creating a self-correcting feedback loop that isolates malicious or unreliable participants. Experiments on a prototype implementation demonstrate 100% detection rate and 0% false positive rate for tampered records, sub-millisecond validation latency for record and hub operations, and effective harness convergence that excludes adversarial nodes within two rounds while promoting honest nodes to trusted status within five rounds.
Figures & tables
Fig. 1: Overview of the HadAgent system architecture. From top to bottom: the Agent Interaction Layer handles inference requests between agents; the Two-Tier Nodes Layer organizes master and secondary nodes with cross-node verification; the Network and Harness Layer provides communication, monitoring, and fault tolerance; and the Block layer stores validated records across three Merkle-rooted lanes (DATA, MODEL, PROOF).
Fig. 2: Overview of Block Design. Each block contains a header and a three-lane body. The header stores the block height, previous block hash, timestamp, and three independent Merkle roots corresponding to the DATA, MODEL, and PROOF lanes. The Data Lane records dataset and audit hashes, the Model Lane records model hashes and version identifiers, and the Proof Lane records input/output hashes and validation scores; all lanes include digital signatures for authentication. Only cryptographic hashes, metadata, validation scores, and signatures are stored on-chain; raw datasets, model weights, and full inference outputs remain off-chain and are referenced by their content hashes.
Fig. 3: Consensus Flow with Interval Verification. The cycle proceeds in eight steps: (1) an agent submits an inference request; (2) a secondary node executes the LLM inference task; (3) the master node evaluates the returned result and computes a validation score; (4) the validated record is queued in the mempool; (5) a cross-master verification interval is triggered, in which master nodes independently re-evaluate a randomly selected subset of tasks drawn from standardized benchmarks (e.g., MMLU, HellaSwag); (6) multiple master nodes aggregate their verification outcomes through multi-master voting; (7) the agreed-upon records are committed to the chain as a finalized block; and (8) the validated response is returned to the requesting agent.
Fig. 4: Demonstration of the Two-Tier Node Architecture and Inference Paths. Left (Optimistic Execution): When a trusted secondary node handles a request, the master node routes the task, the trusted node runs inference and returns the result directly to the agent without waiting for consensus, enabling real-time serving. The master node evaluates the result and enters it into the mempool in the background. Right (Verified Execution): When a new (non-trusted) secondary node handles a request, its result is not delivered immediately. Instead, the submitted result must pass master evaluation, enter the mempool, and undergo full consensus verification with a retroactive check before the response is delivered to the agent.
Large Reasoning Models (LRMs) and Multi-Agent Systems (MAS) in high-stakes domains demand reliable verification, yet centralized approaches suffer four limitations: (1) Robustness, with single points of failure vulnerable to attacks and bias; (2) Scalability, as reasoning complexity creates bottlenecks; (3) Opacity, as hidden auditing erodes trust; and (4) Privacy, as exposed reasoning traces risk model theft. We introduce TRUST (Transparent, Robust, and Unified Services for Trustworthy AI), a decentralized framework with three innovations: (i) Hierarchical Directed Acyclic Graphs (HDAGs) that decompose Chain-of-Thought reasoning into five abstraction levels for parallel distributed auditing; (ii) the DAAN protocol, which projects multi-agent interactions into Causal Interaction Graphs (CIGs) for deterministic root-cause attribution; and (iii) a multi-tier consensus mechanism among computational checkers, LLM evaluators, and human experts with stake-weighted voting that guarantees correctness under 30% adversarial participation. We prove a Safety-Profitability Theorem ensuring honest auditors profit while malicious actors incur losses. All decisions are recorded on-chain, while privacy-by-design segmentation prevents reconstruction of proprietary logic. Across multiple LLMs and benchmarks, TRUST attains 72.4% accuracy (4-18% above baselines) and remains resilient against 20% corruption. DAAN reaches 70% root-cause attribution (vs. 54-63% for standard methods) with 60% token savings. Human studies validate the design (F1 = 0.89, Brier = 0.074). The framework supports (A1) decentralized auditing, (A2) tamper-proof leaderboards, (A3) trustless data annotation, and (A4) governed autonomous agents, pioneering decentralized AI auditing for safe, accountable deployment of reasoning-capable systems.
Yu-Chao Huang, Zhen Tan, Mohan Zhang +3
UNITES Lab, University of North Carolina at Chapel Hill · Arizona State University · Columbia University
General-purpose large language model agents have achieved strong performance on tool-augmented tasks, yet they rely on assumptions break down in blockchain environments. On-chain execution is stateful, adversarial, and economically irreversible, exposing three fundamental gaps: Reactivity, Irreversibility, and Observability. We propose ChainClaw, a blockchain-native agent framework built on OpenClaw, that addresses all three gaps through a layered architecture comprising an event-driven orchestration layer, a simulation-based safety intelligence layer, and an on-chain monitoring runtime layer, unified by a cross-layer memory subsystem. ChainClaw closes the Reactivity gap via event ingestion and simulation feedback, the Irreversibility gap via a pre-execution safety pipeline with transaction simulation and action guard, and the Observability gap via an on-chain read adapter and transaction monitor. We evaluate ChainClaw on a purpose-built benchmark covering seven tasks across four categories and five dimensions. ChainClaw consistently outperforms representative baselines on both safety and task completion.
Jiacheng Wei, Zhaoxin Fan, Xin Wen +5
Beijing Advanced Innovation Center for Future Blockchain and Privacy Computing, Beihang University, Beijing, China · School of Artificial Intelligence, Beihang University, Beijing, China · Zhongguancun Laboratory, Beihang University, Beijing, China +1
Advances in Artificial Intelligence (AI) have led AI for Theorem Proving to become a promising means of formally verifying computer systems. Whilst formal verification is traditionally reserved for safety-critical systems due to the required amount of expertise and effort, AI can help to automate a large amount of this workload and make it far more accessible. Blockchain-based systems are becoming increasingly popular and are frequently targeted by malicious actors, often resulting in huge financial losses, highlighting the need to better verify these systems and mitigate vulnerabilities. Arguably the most important component of these systems is the consensus protocol, which allows nodes to agree on decisions in a potentially adversarial environment. In this paper, we improve upon IsabeLLM, the automated theorem proving tool in Isabelle. Namely, we implement a Retrieval-Augmented Generation framework, Error tracing and counterexample generation for improved context supplied to the Large Language Model. Compatibility with the latest version of Isabelle and Sledgehammer is also implemented for improved efficiency. We compare the performance of the two versions of IsabeLLM in their ability to complete the verification of Bitcoin's Proof of Work consensus.